PyTorch_Stock_Returns.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "PyTorch Stock Returns.ipynb",
      "provenance": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/delhub/public-temp/blob/master/PyTorch_Stock_Returns.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3YERLTJjD4kk",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import torch\n",
        "import torch.nn as nn\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "from sklearn.preprocessing import StandardScaler"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7JQ-aWBwH6zM",
        "colab_type": "text"
      },
      "source": [
        "# Part 1: Price Prediction"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0ARJutSvD772",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# yes, you can read dataframes from URLs!\n",
        "df = pd.read_csv('https://raw.githubusercontent.com/lazyprogrammer/machine_learning_examples/master/tf2.0/sbux.csv')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2Enitd48EA1e",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        },
        "outputId": "44eca8ef-44d5-4631-d895-e1ad61ee4fc2"
      },
      "source": [
        "df.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>date</th>\n",
              "      <th>open</th>\n",
              "      <th>high</th>\n",
              "      <th>low</th>\n",
              "      <th>close</th>\n",
              "      <th>volume</th>\n",
              "      <th>Name</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>2013-02-08</td>\n",
              "      <td>27.920</td>\n",
              "      <td>28.325</td>\n",
              "      <td>27.920</td>\n",
              "      <td>28.185</td>\n",
              "      <td>7146296</td>\n",
              "      <td>SBUX</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>2013-02-11</td>\n",
              "      <td>28.260</td>\n",
              "      <td>28.260</td>\n",
              "      <td>27.930</td>\n",
              "      <td>28.070</td>\n",
              "      <td>5457354</td>\n",
              "      <td>SBUX</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>2013-02-12</td>\n",
              "      <td>28.000</td>\n",
              "      <td>28.275</td>\n",
              "      <td>27.975</td>\n",
              "      <td>28.130</td>\n",
              "      <td>8665592</td>\n",
              "      <td>SBUX</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>2013-02-13</td>\n",
              "      <td>28.230</td>\n",
              "      <td>28.230</td>\n",
              "      <td>27.750</td>\n",
              "      <td>27.915</td>\n",
              "      <td>7022056</td>\n",
              "      <td>SBUX</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>2013-02-14</td>\n",
              "      <td>27.765</td>\n",
              "      <td>27.905</td>\n",
              "      <td>27.675</td>\n",
              "      <td>27.775</td>\n",
              "      <td>8899188</td>\n",
              "      <td>SBUX</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "         date    open    high     low   close   volume  Name\n",
              "0  2013-02-08  27.920  28.325  27.920  28.185  7146296  SBUX\n",
              "1  2013-02-11  28.260  28.260  27.930  28.070  5457354  SBUX\n",
              "2  2013-02-12  28.000  28.275  27.975  28.130  8665592  SBUX\n",
              "3  2013-02-13  28.230  28.230  27.750  27.915  7022056  SBUX\n",
              "4  2013-02-14  27.765  27.905  27.675  27.775  8899188  SBUX"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kCITTItVECqS",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        },
        "outputId": "dade9bc9-6c6e-4b63-910d-2f455172b755"
      },
      "source": [
        "df.tail()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>date</th>\n",
              "      <th>open</th>\n",
              "      <th>high</th>\n",
              "      <th>low</th>\n",
              "      <th>close</th>\n",
              "      <th>volume</th>\n",
              "      <th>Name</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>1254</th>\n",
              "      <td>2018-02-01</td>\n",
              "      <td>56.280</td>\n",
              "      <td>56.42</td>\n",
              "      <td>55.89</td>\n",
              "      <td>56.00</td>\n",
              "      <td>14690146</td>\n",
              "      <td>SBUX</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1255</th>\n",
              "      <td>2018-02-02</td>\n",
              "      <td>55.900</td>\n",
              "      <td>56.32</td>\n",
              "      <td>55.70</td>\n",
              "      <td>55.77</td>\n",
              "      <td>15358909</td>\n",
              "      <td>SBUX</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1256</th>\n",
              "      <td>2018-02-05</td>\n",
              "      <td>55.530</td>\n",
              "      <td>56.26</td>\n",
              "      <td>54.57</td>\n",
              "      <td>54.69</td>\n",
              "      <td>16059955</td>\n",
              "      <td>SBUX</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1257</th>\n",
              "      <td>2018-02-06</td>\n",
              "      <td>53.685</td>\n",
              "      <td>56.06</td>\n",
              "      <td>53.56</td>\n",
              "      <td>55.61</td>\n",
              "      <td>17415065</td>\n",
              "      <td>SBUX</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1258</th>\n",
              "      <td>2018-02-07</td>\n",
              "      <td>55.080</td>\n",
              "      <td>55.43</td>\n",
              "      <td>54.44</td>\n",
              "      <td>54.46</td>\n",
              "      <td>13927022</td>\n",
              "      <td>SBUX</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "            date    open   high    low  close    volume  Name\n",
              "1254  2018-02-01  56.280  56.42  55.89  56.00  14690146  SBUX\n",
              "1255  2018-02-02  55.900  56.32  55.70  55.77  15358909  SBUX\n",
              "1256  2018-02-05  55.530  56.26  54.57  54.69  16059955  SBUX\n",
              "1257  2018-02-06  53.685  56.06  53.56  55.61  17415065  SBUX\n",
              "1258  2018-02-07  55.080  55.43  54.44  54.46  13927022  SBUX"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lvZLvhINEEXm",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Start by doing the WRONG thing - trying to predict the price itself\n",
        "series = df['close'].values.reshape(-1, 1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sJlZEBuhEIhx",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Normalize the data\n",
        "# Note: I didn't think about where the true boundary is, this is just approx.\n",
        "scaler = StandardScaler()\n",
        "scaler.fit(series[:len(series) // 2])\n",
        "series = scaler.transform(series).flatten()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lFWSq_z-EKp6",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "1651f2e6-5250-4b20-bbed-5f89251840a6"
      },
      "source": [
        "### build the dataset\n",
        "# let's see if we can use T past values to predict the next value\n",
        "T = 20\n",
        "D = 1\n",
        "X = []\n",
        "Y = []\n",
        "for t in range(len(series) - T):\n",
        "  x = series[t:t+T]\n",
        "  X.append(x)\n",
        "  y = series[t+T]\n",
        "  Y.append(y)\n",
        "\n",
        "X = np.array(X).reshape(-1, T, 1) # Now the data should be N x T x D\n",
        "Y = np.array(Y).reshape(-1, 1)\n",
        "N = len(X)\n",
        "print(\"X.shape\", X.shape, \"Y.shape\", Y.shape)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "X.shape (1239, 20, 1) Y.shape (1239, 1)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ejA41yIHEMMs",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "### try autoregressive RNN model\n",
        "class RNN(nn.Module):\n",
        "  def __init__(self, n_inputs, n_hidden, n_rnnlayers, n_outputs):\n",
        "    super(RNN, self).__init__()\n",
        "    self.D = n_inputs\n",
        "    self.M = n_hidden\n",
        "    self.K = n_outputs\n",
        "    self.L = n_rnnlayers\n",
        "\n",
        "    self.rnn = nn.LSTM(\n",
        "        input_size=self.D,\n",
        "        hidden_size=self.M,\n",
        "        num_layers=self.L,\n",
        "        batch_first=True)\n",
        "    self.fc = nn.Linear(self.M, self.K)\n",
        "  \n",
        "  def forward(self, X):\n",
        "    # initial hidden states\n",
        "    h0 = torch.zeros(self.L, X.size(0), self.M).to(device)\n",
        "    c0 = torch.zeros(self.L, X.size(0), self.M).to(device)\n",
        "\n",
        "    # get RNN unit output\n",
        "    out, _ = self.rnn(X, (h0, c0))\n",
        "\n",
        "    # we only want h(T) at the final time step\n",
        "    out = self.fc(out[:, -1, :])\n",
        "    return out"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hxjVvws8EUfr",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model = RNN(1, 5, 1, 1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7zL5f7KWEcVw",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 102
        },
        "outputId": "4680ee2a-4550-4651-a833-3c86341da2bc"
      },
      "source": [
        "device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n",
        "print(device)\n",
        "model.to(device)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "cpu\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "RNN(\n",
              "  (rnn): LSTM(1, 5, batch_first=True)\n",
              "  (fc): Linear(in_features=5, out_features=1, bias=True)\n",
              ")"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 10
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "h8Qlpc1sEvcA",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Loss and optimizer\n",
        "criterion = nn.MSELoss()\n",
        "optimizer = torch.optim.Adam(model.parameters(), lr=0.1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SbUEPOnZE1c9",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Training\n",
        "def full_gd(model,\n",
        "            criterion,\n",
        "            optimizer,\n",
        "            X_train,\n",
        "            y_train,\n",
        "            X_test,\n",
        "            y_test,\n",
        "            epochs=200):\n",
        "\n",
        "  # Stuff to store\n",
        "  train_losses = np.zeros(epochs)\n",
        "  test_losses = np.zeros(epochs)\n",
        "\n",
        "  for it in range(epochs):\n",
        "    # zero the parameter gradients\n",
        "    optimizer.zero_grad()\n",
        "\n",
        "    # Forward pass\n",
        "    outputs = model(X_train)\n",
        "    loss = criterion(outputs, y_train)\n",
        "      \n",
        "    # Backward and optimize\n",
        "    loss.backward()\n",
        "    optimizer.step()\n",
        "\n",
        "    # Save losses\n",
        "    train_losses[it] = loss.item()\n",
        "\n",
        "    # Test loss\n",
        "    test_outputs = model(X_test)\n",
        "    test_loss = criterion(test_outputs, y_test)\n",
        "    test_losses[it] = test_loss.item()\n",
        "      \n",
        "    if (it + 1) % 5 == 0:\n",
        "      print(f'Epoch {it+1}/{epochs}, Train Loss: {loss.item():.4f}, Test Loss: {test_loss.item():.4f}')\n",
        "  \n",
        "  return train_losses, test_losses"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lnQkAXuxE8G5",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Make inputs and targets\n",
        "X_train = torch.from_numpy(X[:-N//2].astype(np.float32))\n",
        "y_train = torch.from_numpy(Y[:-N//2].astype(np.float32))\n",
        "X_test = torch.from_numpy(X[-N//2:].astype(np.float32))\n",
        "y_test = torch.from_numpy(Y[-N//2:].astype(np.float32))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QM8yEi78E_RG",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# move data to GPU\n",
        "X_train, y_train = X_train.to(device), y_train.to(device)\n",
        "X_test, y_test = X_test.to(device), y_test.to(device)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eeU5KgjXFA7S",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 697
        },
        "outputId": "e8438fd3-53ff-441b-c969-cf54563fcfe9"
      },
      "source": [
        "train_losses, test_losses = full_gd(model,\n",
        "                                    criterion,\n",
        "                                    optimizer,\n",
        "                                    X_train,\n",
        "                                    y_train,\n",
        "                                    X_test,\n",
        "                                    y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 5/200, Train Loss: 0.1533, Test Loss: 0.2875\n",
            "Epoch 10/200, Train Loss: 0.0513, Test Loss: 0.2019\n",
            "Epoch 15/200, Train Loss: 0.0236, Test Loss: 0.0842\n",
            "Epoch 20/200, Train Loss: 0.0223, Test Loss: 0.1219\n",
            "Epoch 25/200, Train Loss: 0.0168, Test Loss: 0.0928\n",
            "Epoch 30/200, Train Loss: 0.0136, Test Loss: 0.0910\n",
            "Epoch 35/200, Train Loss: 0.0109, Test Loss: 0.1064\n",
            "Epoch 40/200, Train Loss: 0.0081, Test Loss: 0.0716\n",
            "Epoch 45/200, Train Loss: 0.0062, Test Loss: 0.0750\n",
            "Epoch 50/200, Train Loss: 0.0056, Test Loss: 0.0560\n",
            "Epoch 55/200, Train Loss: 0.0056, Test Loss: 0.0532\n",
            "Epoch 60/200, Train Loss: 0.0055, Test Loss: 0.0493\n",
            "Epoch 65/200, Train Loss: 0.0055, Test Loss: 0.0469\n",
            "Epoch 70/200, Train Loss: 0.0054, Test Loss: 0.0466\n",
            "Epoch 75/200, Train Loss: 0.0053, Test Loss: 0.0444\n",
            "Epoch 80/200, Train Loss: 0.0053, Test Loss: 0.0462\n",
            "Epoch 85/200, Train Loss: 0.0053, Test Loss: 0.0445\n",
            "Epoch 90/200, Train Loss: 0.0052, Test Loss: 0.0460\n",
            "Epoch 95/200, Train Loss: 0.0052, Test Loss: 0.0453\n",
            "Epoch 100/200, Train Loss: 0.0052, Test Loss: 0.0455\n",
            "Epoch 105/200, Train Loss: 0.0052, Test Loss: 0.0446\n",
            "Epoch 110/200, Train Loss: 0.0052, Test Loss: 0.0445\n",
            "Epoch 115/200, Train Loss: 0.0052, Test Loss: 0.0436\n",
            "Epoch 120/200, Train Loss: 0.0052, Test Loss: 0.0432\n",
            "Epoch 125/200, Train Loss: 0.0052, Test Loss: 0.0425\n",
            "Epoch 130/200, Train Loss: 0.0052, Test Loss: 0.0421\n",
            "Epoch 135/200, Train Loss: 0.0052, Test Loss: 0.0415\n",
            "Epoch 140/200, Train Loss: 0.0052, Test Loss: 0.0411\n",
            "Epoch 145/200, Train Loss: 0.0052, Test Loss: 0.0406\n",
            "Epoch 150/200, Train Loss: 0.0052, Test Loss: 0.0401\n",
            "Epoch 155/200, Train Loss: 0.0051, Test Loss: 0.0397\n",
            "Epoch 160/200, Train Loss: 0.0051, Test Loss: 0.0392\n",
            "Epoch 165/200, Train Loss: 0.0051, Test Loss: 0.0387\n",
            "Epoch 170/200, Train Loss: 0.0051, Test Loss: 0.0383\n",
            "Epoch 175/200, Train Loss: 0.0051, Test Loss: 0.0379\n",
            "Epoch 180/200, Train Loss: 0.0051, Test Loss: 0.0374\n",
            "Epoch 185/200, Train Loss: 0.0051, Test Loss: 0.0370\n",
            "Epoch 190/200, Train Loss: 0.0051, Test Loss: 0.0366\n",
            "Epoch 195/200, Train Loss: 0.0051, Test Loss: 0.0362\n",
            "Epoch 200/200, Train Loss: 0.0051, Test Loss: 0.0357\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TA9ZKt4aFIGO",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        },
        "outputId": "a893539f-a28e-4370-b07a-0d757cfae672"
      },
      "source": [
        "# Plot the train loss and test loss per iteration\n",
        "plt.plot(train_losses, label='train loss')\n",
        "plt.plot(test_losses, label='test loss')\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWoAAAD4CAYAAADFAawfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1vAQGe6JFSM9",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# One-step forecast using true targets\n",
        "validation_target = Y\n",
        "validation_predictions = []\n",
        "\n",
        "# index of first validation input\n",
        "i = 0\n",
        "\n",
        "X_on_device = torch.from_numpy(X.astype(np.float32)).to(device)\n",
        "\n",
        "while len(validation_predictions) < len(validation_target):\n",
        "  input_ = X_on_device[i].reshape(1, T, 1)\n",
        "  p = model(input_)[0,0].item() # 1x1 array -> scalar\n",
        "  i += 1\n",
        "  \n",
        "  # update the predictions list\n",
        "  validation_predictions.append(p)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WmlRrKo8FU9d",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "outputId": "dd45ebf3-7c3f-4e9b-dff2-9503d77f1cca"
      },
      "source": [
        "plt.plot(validation_target, label='forecast target')\n",
        "plt.plot(validation_predictions, label='forecast prediction')\n",
        "plt.legend()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7f05b4820e80>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 18
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXIAAAD4CAYAAADxeG0DAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nhcwphbxQKBQypxTyQqFQyJw/ASk7TmXadgHzAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "o_nIYq4tGU4g",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Multi-step forecast\n",
        "validation_target = Y[-N//2:]\n",
        "validation_predictions = []\n",
        "\n",
        "# last train input\n",
        "last_x = X_test[0].view(T)\n",
        "\n",
        "while len(validation_predictions) < len(validation_target):\n",
        "  input_ = last_x.reshape(1, T, 1)\n",
        "  p = model(input_)\n",
        "  # [0,0] # 1x1 array -> scalar\n",
        "  \n",
        "  # update the predictions list\n",
        "  validation_predictions.append(p[0,0].item())\n",
        "  \n",
        "  # make the new input\n",
        "  last_x = torch.cat((last_x[1:], p[0]))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rc-qBmQlGs7-",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "outputId": "f59c596a-920c-4f80-cd2b-8b19f7abfba4"
      },
      "source": [
        "plt.plot(validation_target, label='forecast target')\n",
        "plt.plot(validation_predictions, label='forecast prediction')\n",
        "plt.legend()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7f05b479d470>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 20
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAD4CAYAAAD8Zh1EAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tHOwz6LxH4RJ",
        "colab_type": "text"
      },
      "source": [
        "# Part 2: Returns"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xBjrWdZmG0KL",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# calculate returns by first shifting the data\n",
        "df['PrevClose'] = df['close'].shift(1) # move everything up 1\n",
        "\n",
        "# so now it's like\n",
        "# close / prev close\n",
        "# x[2] x[1]\n",
        "# x[3] x[2]\n",
        "# x[4] x[3]\n",
        "# ...\n",
        "# x[t] x[t-1]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "a_JMC6A9G2Sr",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        },
        "outputId": "401c7d17-b431-4ef8-958f-5cd28238b5eb"
      },
      "source": [
        "df.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>date</th>\n",
              "      <th>open</th>\n",
              "      <th>high</th>\n",
              "      <th>low</th>\n",
              "      <th>close</th>\n",
              "      <th>volume</th>\n",
              "      <th>Name</th>\n",
              "      <th>PrevClose</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>2013-02-08</td>\n",
              "      <td>27.920</td>\n",
              "      <td>28.325</td>\n",
              "      <td>27.920</td>\n",
              "      <td>28.185</td>\n",
              "      <td>7146296</td>\n",
              "      <td>SBUX</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>2013-02-11</td>\n",
              "      <td>28.260</td>\n",
              "      <td>28.260</td>\n",
              "      <td>27.930</td>\n",
              "      <td>28.070</td>\n",
              "      <td>5457354</td>\n",
              "      <td>SBUX</td>\n",
              "      <td>28.185</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>2013-02-12</td>\n",
              "      <td>28.000</td>\n",
              "      <td>28.275</td>\n",
              "      <td>27.975</td>\n",
              "      <td>28.130</td>\n",
              "      <td>8665592</td>\n",
              "      <td>SBUX</td>\n",
              "      <td>28.070</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>2013-02-13</td>\n",
              "      <td>28.230</td>\n",
              "      <td>28.230</td>\n",
              "      <td>27.750</td>\n",
              "      <td>27.915</td>\n",
              "      <td>7022056</td>\n",
              "      <td>SBUX</td>\n",
              "      <td>28.130</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>2013-02-14</td>\n",
              "      <td>27.765</td>\n",
              "      <td>27.905</td>\n",
              "      <td>27.675</td>\n",
              "      <td>27.775</td>\n",
              "      <td>8899188</td>\n",
              "      <td>SBUX</td>\n",
              "      <td>27.915</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "         date    open    high     low   close   volume  Name  PrevClose\n",
              "0  2013-02-08  27.920  28.325  27.920  28.185  7146296  SBUX        NaN\n",
              "1  2013-02-11  28.260  28.260  27.930  28.070  5457354  SBUX     28.185\n",
              "2  2013-02-12  28.000  28.275  27.975  28.130  8665592  SBUX     28.070\n",
              "3  2013-02-13  28.230  28.230  27.750  27.915  7022056  SBUX     28.130\n",
              "4  2013-02-14  27.765  27.905  27.675  27.775  8899188  SBUX     27.915"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 22
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PynKcwMFG4Ur",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# then the return is\n",
        "# (x[t] - x[t-1]) / x[t-1]\n",
        "df['Return'] = (df['close'] - df['PrevClose']) / df['PrevClose']"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vNkf8pYWG6gK",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        },
        "outputId": "6bf74e2c-7a1a-4156-bc1f-50b681356ca3"
      },
      "source": [
        "df.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>date</th>\n",
              "      <th>open</th>\n",
              "      <th>high</th>\n",
              "      <th>low</th>\n",
              "      <th>close</th>\n",
              "      <th>volume</th>\n",
              "      <th>Name</th>\n",
              "      <th>PrevClose</th>\n",
              "      <th>Return</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>2013-02-08</td>\n",
              "      <td>27.920</td>\n",
              "      <td>28.325</td>\n",
              "      <td>27.920</td>\n",
              "      <td>28.185</td>\n",
              "      <td>7146296</td>\n",
              "      <td>SBUX</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>2013-02-11</td>\n",
              "      <td>28.260</td>\n",
              "      <td>28.260</td>\n",
              "      <td>27.930</td>\n",
              "      <td>28.070</td>\n",
              "      <td>5457354</td>\n",
              "      <td>SBUX</td>\n",
              "      <td>28.185</td>\n",
              "      <td>-0.004080</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>2013-02-12</td>\n",
              "      <td>28.000</td>\n",
              "      <td>28.275</td>\n",
              "      <td>27.975</td>\n",
              "      <td>28.130</td>\n",
              "      <td>8665592</td>\n",
              "      <td>SBUX</td>\n",
              "      <td>28.070</td>\n",
              "      <td>0.002138</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>2013-02-13</td>\n",
              "      <td>28.230</td>\n",
              "      <td>28.230</td>\n",
              "      <td>27.750</td>\n",
              "      <td>27.915</td>\n",
              "      <td>7022056</td>\n",
              "      <td>SBUX</td>\n",
              "      <td>28.130</td>\n",
              "      <td>-0.007643</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>2013-02-14</td>\n",
              "      <td>27.765</td>\n",
              "      <td>27.905</td>\n",
              "      <td>27.675</td>\n",
              "      <td>27.775</td>\n",
              "      <td>8899188</td>\n",
              "      <td>SBUX</td>\n",
              "      <td>27.915</td>\n",
              "      <td>-0.005015</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "         date    open    high     low  ...   volume  Name PrevClose    Return\n",
              "0  2013-02-08  27.920  28.325  27.920  ...  7146296  SBUX       NaN       NaN\n",
              "1  2013-02-11  28.260  28.260  27.930  ...  5457354  SBUX    28.185 -0.004080\n",
              "2  2013-02-12  28.000  28.275  27.975  ...  8665592  SBUX    28.070  0.002138\n",
              "3  2013-02-13  28.230  28.230  27.750  ...  7022056  SBUX    28.130 -0.007643\n",
              "4  2013-02-14  27.765  27.905  27.675  ...  8899188  SBUX    27.915 -0.005015\n",
              "\n",
              "[5 rows x 9 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 24
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LXD_mqvRoaP5",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        },
        "outputId": "18a7a344-2a1e-467f-f111-c9d1510e350b"
      },
      "source": [
        "plt.plot(df['Return']);"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYkAAAD4CAYAAAAZ1BptAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ePCjxG3WG8O5",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "outputId": "de8d11a0-2394-4bb7-d50b-8648033c8048"
      },
      "source": [
        "# Now let's try an LSTM to predict returns\n",
        "df['Return'].hist()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7f05b34c6d30>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 26
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD4CAYAAAAXUaZHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7bpSyngPG-4d",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "series = df['Return'].values[1:].reshape(-1, 1)\n",
        "\n",
        "# Normalize the data\n",
        "# Note: I didn't think about where the true boundary is, this is just approx.\n",
        "scaler = StandardScaler()\n",
        "scaler.fit(series[:len(series) // 2])\n",
        "series = scaler.transform(series).flatten()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "l2WJxImAHBdc",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "39ec057e-3c4c-4477-f149-869ea8886834"
      },
      "source": [
        "### build the dataset\n",
        "# let's see if we can use T past values to predict the next value\n",
        "T = 20\n",
        "D = 1\n",
        "X = []\n",
        "Y = []\n",
        "for t in range(len(series) - T):\n",
        "  x = series[t:t+T]\n",
        "  X.append(x)\n",
        "  y = series[t+T]\n",
        "  Y.append(y)\n",
        "\n",
        "X = np.array(X).reshape(-1, T, 1) # Now the data should be N x T x D\n",
        "Y = np.array(Y).reshape(-1, 1)\n",
        "N = len(X)\n",
        "print(\"X.shape\", X.shape, \"Y.shape\", Y.shape)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "X.shape (1238, 20, 1) Y.shape (1238, 1)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HAwpOxy_HF_6",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 85
        },
        "outputId": "15952d6e-45bf-4def-90fb-103329560ae6"
      },
      "source": [
        "### try autoregressive RNN model\n",
        "model = RNN(1, 5, 1, 1)\n",
        "model.to(device)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "RNN(\n",
              "  (rnn): LSTM(1, 5, batch_first=True)\n",
              "  (fc): Linear(in_features=5, out_features=1, bias=True)\n",
              ")"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 29
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iR1GzyxHHNp6",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Loss and optimizer\n",
        "criterion = nn.MSELoss()\n",
        "optimizer = torch.optim.Adam(model.parameters(), lr=0.01)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "h7cIJyaWHSoM",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Make inputs and targets\n",
        "X_train = torch.from_numpy(X[:-N//2].astype(np.float32))\n",
        "y_train = torch.from_numpy(Y[:-N//2].astype(np.float32))\n",
        "X_test = torch.from_numpy(X[-N//2:].astype(np.float32))\n",
        "y_test = torch.from_numpy(Y[-N//2:].astype(np.float32))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YDapFNV3HVJ7",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# move data to GPU\n",
        "X_train, y_train = X_train.to(device), y_train.to(device)\n",
        "X_test, y_test = X_test.to(device), y_test.to(device)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WEgfpWY8HXSh",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 697
        },
        "outputId": "b5d88d6a-6951-4a11-ac19-e700ffd259cf"
      },
      "source": [
        "train_losses, test_losses = full_gd(model,\n",
        "                                    criterion,\n",
        "                                    optimizer,\n",
        "                                    X_train,\n",
        "                                    y_train,\n",
        "                                    X_test,\n",
        "                                    y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 5/200, Train Loss: 1.0666, Test Loss: 1.0724\n",
            "Epoch 10/200, Train Loss: 1.0527, Test Loss: 1.0909\n",
            "Epoch 15/200, Train Loss: 1.0557, Test Loss: 1.0907\n",
            "Epoch 20/200, Train Loss: 1.0484, Test Loss: 1.0770\n",
            "Epoch 25/200, Train Loss: 1.0476, Test Loss: 1.0735\n",
            "Epoch 30/200, Train Loss: 1.0461, Test Loss: 1.0746\n",
            "Epoch 35/200, Train Loss: 1.0427, Test Loss: 1.0791\n",
            "Epoch 40/200, Train Loss: 1.0401, Test Loss: 1.0858\n",
            "Epoch 45/200, Train Loss: 1.0374, Test Loss: 1.0897\n",
            "Epoch 50/200, Train Loss: 1.0346, Test Loss: 1.0920\n",
            "Epoch 55/200, Train Loss: 1.0327, Test Loss: 1.0969\n",
            "Epoch 60/200, Train Loss: 1.0312, Test Loss: 1.1038\n",
            "Epoch 65/200, Train Loss: 1.0300, Test Loss: 1.1083\n",
            "Epoch 70/200, Train Loss: 1.0286, Test Loss: 1.1085\n",
            "Epoch 75/200, Train Loss: 1.0272, Test Loss: 1.1080\n",
            "Epoch 80/200, Train Loss: 1.0258, Test Loss: 1.1086\n",
            "Epoch 85/200, Train Loss: 1.0244, Test Loss: 1.1095\n",
            "Epoch 90/200, Train Loss: 1.0229, Test Loss: 1.1113\n",
            "Epoch 95/200, Train Loss: 1.0212, Test Loss: 1.1132\n",
            "Epoch 100/200, Train Loss: 1.0192, Test Loss: 1.1146\n",
            "Epoch 105/200, Train Loss: 1.0170, Test Loss: 1.1161\n",
            "Epoch 110/200, Train Loss: 1.0143, Test Loss: 1.1165\n",
            "Epoch 115/200, Train Loss: 1.0112, Test Loss: 1.1174\n",
            "Epoch 120/200, Train Loss: 1.0075, Test Loss: 1.1185\n",
            "Epoch 125/200, Train Loss: 1.0030, Test Loss: 1.1208\n",
            "Epoch 130/200, Train Loss: 0.9975, Test Loss: 1.1247\n",
            "Epoch 135/200, Train Loss: 0.9909, Test Loss: 1.1320\n",
            "Epoch 140/200, Train Loss: 0.9839, Test Loss: 1.1464\n",
            "Epoch 145/200, Train Loss: 0.9769, Test Loss: 1.1659\n",
            "Epoch 150/200, Train Loss: 0.9691, Test Loss: 1.1811\n",
            "Epoch 155/200, Train Loss: 0.9615, Test Loss: 1.1960\n",
            "Epoch 160/200, Train Loss: 0.9546, Test Loss: 1.2168\n",
            "Epoch 165/200, Train Loss: 0.9477, Test Loss: 1.2387\n",
            "Epoch 170/200, Train Loss: 0.9400, Test Loss: 1.2484\n",
            "Epoch 175/200, Train Loss: 0.9316, Test Loss: 1.2681\n",
            "Epoch 180/200, Train Loss: 0.9193, Test Loss: 1.3109\n",
            "Epoch 185/200, Train Loss: 0.9037, Test Loss: 1.3517\n",
            "Epoch 190/200, Train Loss: 0.8904, Test Loss: 1.4092\n",
            "Epoch 195/200, Train Loss: 0.8778, Test Loss: 1.4661\n",
            "Epoch 200/200, Train Loss: 0.8682, Test Loss: 1.4862\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4kmEBf1DHbIh",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        },
        "outputId": "1cefc3ca-aa73-4f07-948d-511ed5171475"
      },
      "source": [
        "# Plot the train loss and test loss per iteration\n",
        "plt.plot(train_losses, label='train loss')\n",
        "plt.plot(test_losses, label='test loss')\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAD4CAYAAAD8Zh1EAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Fkr8I84dHip4",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# One-step forecast using true targets\n",
        "validation_target = Y\n",
        "validation_predictions = []\n",
        "\n",
        "# index of first validation input\n",
        "i = 0\n",
        "\n",
        "X_on_device = torch.from_numpy(X.astype(np.float32)).to(device)\n",
        "\n",
        "while len(validation_predictions) < len(validation_target):\n",
        "  input_ = X_on_device[i].reshape(1, T, 1)\n",
        "  p = model(input_)[0,0].item() # 1x1 array -> scalar\n",
        "  i += 1\n",
        "  \n",
        "  # update the predictions list\n",
        "  validation_predictions.append(p)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pcLxuoUkHk1M",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "outputId": "5160afca-bb4a-4ffe-d39d-1d042d23a6a3"
      },
      "source": [
        "plt.plot(validation_target, label='forecast target')\n",
        "plt.plot(validation_predictions, label='forecast prediction')\n",
        "plt.legend()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7f05b3383e48>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 36
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXIAAAD4CAYAAADxeG0DAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nmVnBOZCbmRWcA7mZWcH9P21//wNhXzYrAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XvoMGhXrHpuT",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Multi-step forecast\n",
        "validation_target = Y[-N//2:]\n",
        "validation_predictions = []\n",
        "\n",
        "# last train input\n",
        "last_x = X_test[0].view(T)\n",
        "\n",
        "while len(validation_predictions) < len(validation_target):\n",
        "  input_ = last_x.reshape(1, T, 1)\n",
        "  p = model(input_)\n",
        "  # [0,0] # 1x1 array -> scalar\n",
        "  \n",
        "  # update the predictions list\n",
        "  validation_predictions.append(p[0,0].item())\n",
        "  \n",
        "  # make the new input\n",
        "  last_x = torch.cat((last_x[1:], p[0]))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WnYmGBvwHsqs",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "outputId": "5f69d066-8732-44ae-c85d-4fee7b8cbc5d"
      },
      "source": [
        "plt.plot(validation_target, label='forecast target')\n",
        "plt.plot(validation_predictions, label='forecast prediction')\n",
        "plt.legend()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7f05b32f4fd0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 38
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXIAAAD4CAYAAADxeG0DAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0\ndHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAgAElEQVR4nOydd5xVxdnHv3POuW07sIs0FRUF6SIq\nihob0agxxhLzxtjyviZqYmKKxhhNiDFRE2OLvURsiQVr7CI2LCAI0qsssvTt9ZZzzrx/nH7LFlhk\nV+/v81H23nvKnDkzv/nN8zzzjJBSkkceeeSRR++FsqsLkEceeeSRx44hT+R55JFHHr0ceSLPI488\n8ujlyBN5HnnkkUcvR57I88gjjzx6ObRdcdPy8nI5dOjQXXHrPPLII49ei3nz5lVLKSvSv98lRD50\n6FDmzp27K26dRx555NFrIYRYl+37vGkljzzyyKOXI0/keeSRRx69HHkizyOPPPLo5dglNvI88sgj\nN1KpFFVVVcTj8V1dlDx2EaLRKEOGDCEUCnXq+DyR55FHD0NVVRXFxcUMHToUIcSuLk4eXzKklNTU\n1FBVVcVee+3VqXPyppU88uhhiMfj9OvXL0/iX1MIIejXr1+XZmR5Is8jjx6IPIl/vdHV99/riPzF\nzzbS0Jba1cXII4888ugx6FVEvra6hZ//Zz6XPTF/Vxcljzy+0rj99tvZf//9Ofvss3d1UQBYsGAB\nr7zySpd/625MmzaNjRs3fin36gp6FZEndROADfVtu7gkeeTx1cZdd93Fm2++yeOPP96p43Vd36nl\n6W4i397y5om8G6DYZiPDzO9qlEceOwsXXXQRn3/+Od/61re45ZZbqK2t5dRTT2Xs2LFMmjSJhQsX\nAjB16lTOOeccJk+ezDnnnINhGFx++eUcdNBBjB07lnvvvde95o033siYMWMYN24cV155JQD3338/\nBx10EOPGjeP000+ntbUVgKeffprRo0czbtw4jjzySJLJJH/4wx948sknGT9+PE8++aR73Wy/zZkz\nh0MPPZQDDjiAww47jBUrVgAWCZ9yyikcc8wxHHvssZimySWXXMKIESOYMmUKJ554ItOnTwdg3rx5\nfOMb3+DAAw/k+OOPZ9OmTUyfPp25c+dy9tlnM378eNraeo6g7FXhh4rN5Hkez+Prgj/9dwlLNzZ2\n6zVHDirhj98elfP3e+65h9dee423336b8vJyLr30Ug444ACef/55Zs6cybnnnsuCBQsAWLp0KbNm\nzSIWi3HfffdRWlrKJ598QiKRYPLkyXzzm99k+fLlvPDCC8yePZuCggJqa2sBOO2007jwwgsBuPrq\nq3nwwQe59NJLufbaa3n99dcZPHgw9fX1hMNhrr32WubOncsdd9wRKGu23xobG3n//ffRNI0ZM2Zw\n1VVX8cwzzwDw6aefsnDhQvr27cv06dOprKxk6dKlbN26lf33358f/ehHpFIpLr30Ul544QUqKip4\n8skn+f3vf8+//vUv7rjjDm666SYmTpzYre9kR9GriFy1Pbl5RZ5HHl8eZs2a5RLhMcccQ01NDY2N\n1uByyimnEIvFAHjjjTdYuHChq2obGhpYtWoVM2bM4IILLqCgoACAvn37ArB48WKuvvpq6uvraW5u\n5vjjjwdg8uTJnH/++Xzve9/jtNNO63J5GxoaOO+881i1ahVCCFIpLzhiypQp7v1nzZrFmWeeiaIo\nDBgwgKOPPhqAFStWsHjxYqZMmQKAYRgMHDiwy+X4MtGriNyBmd8wOo+vCdpTzj0BhYWF7t9SSv75\nz3+6hOzg9ddfz3ru+eefz/PPP8+4ceOYNm0a77zzDmDNCGbPns3LL7/MgQceyLx587pUpmuuuYaj\njz6a5557jsrKSo466qis5c0FKSWjRo3io48+6tJ9dyW6zUYuhFCFEPOFEC911zXT4RC4mVfkeeTx\npeGII45wnZ7vvPMO5eXllJSUZBx3/PHHc/fdd7sKeOXKlbS0tDBlyhQeeugh1wbumFaampoYOHAg\nqVQq4FRds2YNhxxyCNdeey0VFRWsX7+e4uJimpqaspYv/beGhgYGDx4MWHbxXJg8eTLPPPMMpmmy\nZcsWdyAZPnw427Ztc4k8lUqxZMmSrPfqKehOZ+cvgGXdeL0MOPxt5BV5Hnl8aZg6dSrz5s1j7Nix\nXHnllTz88MNZj/u///s/Ro4cyYQJExg9ejQ/+clP0HWdE044gVNOOYWJEycyfvx4brrpJgD+/Oc/\nc8ghhzB58mRGjBjhXufyyy9nzJgxjB49msMOO4xx48Zx9NFHs3Tp0gxnJ5Dx2xVXXMHvfvc7Djjg\ngHajU04//XSGDBnCyJEj+eEPf8iECRMoLS0lHA4zffp0fvvb3zJu3DjGjx/Phx9+CFiziIsuuqjH\nOTuF7AZSFEIMAR4G/gL8Skp5cnvHT5w4UW7PxhIrtzTxzVveo6I4wie/P277CptHHj0cy5YtY//9\n99/VxfhaoLm5maKiImpqajj44IP54IMPGDBgwK4uFpC9HQgh5kkpMzyt3WUjvxW4AijuputlhWNa\n6Y7BJ4888sjj5JNPpr6+nmQyyTXXXNNjSLyr2GEiF0KcDGyVUs4TQhzVznE/Bn4MsMcee2zXvUxr\nPVA+aiWPPPLoFjh28d6O7rCRTwZOEUJUAk8AxwghHks/SEp5n5RyopRyYkVFxt6hnYLr7MzzeB55\n5JGHix0mcinl76SUQ6SUQ4HvAzOllD/c4ZJlQT5qJY888sgjE71qiX4+aiWPPPLIIxPduiBISvkO\n8E53XtMPR5HnbeR55JFHHh56lyI3naiVXVyQPPL4iqM3pbHtbrzzzjucfLIVQf3iiy9yww035Dy2\nvr6eu+66y/28ceNGzjjjjJ1exnT0LiLPm1byyONLQW9KY9sZSCkxnbC3LuCUU05xszVmQzqRDxo0\nyM0182WilxF53rSSRx47G70pjS1Yy/C/853vcNRRR7Hvvvvypz/9CYDKykqGDx/Oueeey+jRo1m/\nfj1vvPEGhx56KBMmTODMM8+kubkZgNdee40RI0YwYcIEnn322cC1f/aznwGwZcsWvvvd7zJu3DjG\njRvHhx9+yJVXXsmaNWsYP348l19+OZWVlYwePRqw9l694IILGDNmDAcccABvv/22e83TTjuNE044\ngX333Zcrrrhih99Zr0qalY9WyeNrh1evhM2LuveaA8bAt3KbC3pTGlsHc+bMYfHixRQUFHDQQQdx\n0kknUV5ezqpVq3j44YeZNGkS1dXVXHfddcyYMYPCwkJuvPFGbr75Zq644gouvPBCZs6cybBhwzjr\nrLOy3uPnP/853/jGN3juuecwDIPm5mZuuOEGFi9e7NZHZWWle/ydd96JEIJFixaxfPlyvvnNb7Jy\n5UrAmmHMnz+fSCTC8OHDufTSS9l999279h596GWKfFeXII88vn6YNWsW55xzDtBxGttHHnmE8ePH\nc8ghh1BTU9NhGtsjjjiCMWPG8Pjjj7uJqZw0tvfffz+GYXSqjFOmTKFfv37EYjFOO+00Zs2aBcCe\ne+7JpEmTAPj4449ZunQpkydPZvz48Tz88MOsW7eO5cuXs9dee7HvvvsihOCHP8wePT1z5kwuvvhi\nAFRVpbS0tMN6c641YsQI9txzT5fIjz32WEpLS4lGo4wcOZJ169Z16jlzoXcp8rxtPI+vG9pRzj0B\nPSWNbfqu887n9PJNmTKF//znP4FjHTX9ZSISibh/q6q6wz6GXqXI807OPPL48tHT09gCvPnmm9TW\n1tLW1sbzzz/P5MmTM46ZNGkSH3zwAatXrwagpaWFlStXMmLECCorK1mzZg1ABtE7OPbYY7n77rsB\na7OJhoaGdsvlr7eVK1fyxRdfMHz48JzPsCPoVUSeT5aVRx5fPnp6GluAgw8+mNNPP52xY8dy+umn\nZ92KraKigmnTpvE///M/jB07lkMPPZTly5cTjUa57777OOmkk5gwYQL9+/fP+ny33XYbb7/9NmPG\njOHAAw9k6dKl9OvXj8mTJzN69Gguv/zywPGXXHIJpmkyZswYzjrrLKZNmxZQ4t2Jbklj21Vsbxrb\nGUu38H+PWOdV3nBSdxcrjzx6BPJpbLuGadOmtesI7a3oShrbXqXI8zbyPPLII49M5Il8O3HbjFWc\ncfeH23Xui59tZOiVL7O+trWbS5VHHl8/nH/++V85Nd5V9LKolV1dAg+3zFi53ee+uGAjAMs2NbJ7\n34LuKlIeXyFIKTMiMfL4+qCrJu+8It8FcPpnTxqY8ug5iEaj1NTU5J37X1NIKampqSEajXb6nF6l\nyL8qS/MVV2h9NZ7Hj9mf11DfluL4Ub1zy6yegCFDhlBVVcW2bdt2dVHy2EWIRqMMGTKk08f3KiL/\nqggUgcXkX5FxKYCz7vsYyEcV7QhCoRB77bXXri5GHr0Ivda0smJzE++s2LoLS7P9UOxa/6oMTL0d\n0+dVMfvzml1djDzy2G70KkXuN60cf+t7QO9Ufp4izzN5T8Bvnv4M6J1tKY88oJcp8q8K7znOzq4+\nzqMfVbKxvq27i5NHHnn0cvQqIv+qKFgnrKwrUQlbm+Jc88ISLnjok51VrDzyyKOXopcR+a4uQffA\niVrpyrjkmJXq25I7oUR55JFHb0avIvKvSvZDJ/qwKzMM51BBfpFIHnnkEUSvIvKeuECis2VK6AZP\nfvIFUkoU17QCx9/yHlNfXLIzi5hHO/iqrE3I4+uNXhW10hO3ejMlqJ0QyTe/uZJ73/2c0ljYleSm\nlKzY0sSKLU1MPWXUzi1oHlmRMrq+IW8eefQ07LAiF0LsLoR4WwixVAixRAjxi+4oWDYYPY/HO63o\nqmqtaJOUYXqKvAv3cY7Np9/oXiT0PJHvbOiGyZ1vryae6ty2bXl0Hd1hWtGBX0spRwKTgJ8KIUZ2\nw3Uz0BNNK521c7cmra2cCsKq6+zsygyjJ85GvgpI5ol8p2P6vCr+/voK/jlz1a4uylcWO0zkUspN\nUspP7b+bgGXA4B29bjb0xPDDzhO5pUZiYdVV5H5u7kjZO/fpyYK8N9qb86YVC68v2czQK19ma2O8\n26/dYrd9pw/0BjTGU71qBtGtzk4hxFDgAGB2d17XQU/sc50lrza7UYRVxYsj9xlXalvaDyvsDRzZ\nG9VtbyzzzsBjH1u7uC/bnHtfzO2FM5NW2rELGqZkycaGbr/39mLs1Dc4457t229gV6DbiFwIUQQ8\nA1wmpWzM8vuPhRBzhRBztzerW89U5J07zlEjkuxpbDsa/XuD2vWTYk80g2VDsieqg68YvNDZ3Pjn\nzFWcdPssFm/oOWS+eEMGjfVYdAuRCyFCWCT+uJTy2WzHSCnvk1JOlFJOrKio2K779ERy6KztujVh\n2cil9KWx9T1PR4TimlZ6sLczoXuDUS8Yd4CuKfLGeIoVO0Gx9iTsjNbltF1FyX31+V/UA7CtKbET\nSvDVR3dErQjgQWCZlPLmHS9SbvREcujsIqVWW3GbUmZNY9sRoXRVkdc0f/kdwh8B0htmENA1RX7O\nA7PdZG3dgcZ4qtuutaPYmRrJaQrtaZDOkH0eudEdinwycA5wjBBigf3fid1w3Qz0RHLotCJPekTu\nLdH3zk13um2sb+P5+Rvcz1159k8qaznwuhm8umhTp89pF2vehpbqDg/rlUTeBUX+WZU17e+OmeGL\nn21k7NQ3eowpwfHX7IwJn3PtjmzkAGoPnnH2ZHRH1MosKaWQUo6VUo63/3ulOwqX5V5Zv/twdTUv\nfrZxZ9wSgM0N8ZzRDZ3lK4cwpPTMI+0p8u/f9zGXPbmAmcu3UN2c6JJ/YKFNOLPX1nb6nJwwDXj0\nVHjkOx0eqps+IrfLG08Zu8b7n2zp3GHb4ezsjjHqg1XWwLiohxC5g52RAqIzNnLddBR5t98+gD+8\nsJgrn1nYtZOqV8Gbf+zR6Vd71RL9bB3IlPCDB2bz8//Mz/gtZZj8e/YXO6QO4ymDSde/xZXPLMr6\ne1fzv1hEbv/t+z59ir++rhWAH02by3n/mtMl8pCy8+rqg9XVrN7ajt3XtGz7bFnc4bX89WzYq7fG\n/ekN9v/Dax0XpCvYshQ+eyL375sXwV8HwZLnOrzU9hB5d8w2NHs5sJ4uEJa+AFNLIf7lOtp2qmnF\n7FiRO8doO5nJH/loHU98sr5rJ/37e/DBrdDQxfO+RPQqIs9Gmu11qoc+WMtVzy3iiU++2O57Ompy\nxrItWX/v6kKd5ZsbXdXjn2GkE4r/UddWt7jP2ZWZZ2fU1dkPzOa4m9ux+5qdV9P+MjvvKqGbSAm/\nfuqzTl+nQ9x9KDz3k9y/b7YH3RUdDyDbE0feHUQeUhX7/mnXev8f1r+1a3b4HtuDnWHZcKqrPfO3\n016GfHYrLM4aL9Ft+LX2FLx1bedPcPqAI2p6IHoVkWczL7Rncmhssyq+umn7U792pFS62qmve3kZ\nn35Rl3HtjA7tQ0RTgs+54VNIxWlO6DtlAUcAsvNEHlDkafXyzKdV3VYkF7lejhKy/u1Ex9ue8EO/\nCSnrNXWTqS8uaXdtgGazWsa1hN0lO7hHd2PnOjs7jrhyBNGgBbfB9At2XmGAS7XnvQGzM1BU6980\nUSOl5O+vL2fVll0fydSriDxbY2uvAXqqJ3enOPgvM7j+lWU5f+/INr09se1VttnEbEeR+xHRVJcY\ny6mF+4+G//6Ck25/n4P/+lbG8a5NsjvUla/xSim58+3VfFHTmv1QmZvIdwqMHETpdryOo0JylfO5\n+VXs/buXAyGVHZ3j4KWFG5n2YSV/f315zmNCWg5Frth57NoZhJ6au55NDZ3bKaoloXcqOsZxSLrv\nsG4dvP3XbmF45wqKEPDFbOu/NPToFNU53kltS5I7317D2Q/slPWPXUKvIvJsHejBWZ/nPD6s5Sby\ny56Yz13vrGZrU4J738t9jY5MGtu3SMm6mL/x+su4uSGosiMhxVUsYWl3yi8+ZF0OQnUjELajZJkX\n88q1tSnB319fwfkPzcl6aCDlwJfRMVM5yEx1FHnHs4lcpHzDq8sxJdQ0Zw4WegdE3mKvGVDbsSXk\nFBnCHoRyzIQa4ymumL6Qcx6cw6KqBv49u32z4aTr32Ls1DfaPcYPtz6eOgfevRFqVnf63FwI+Gz+\n9U3rv4z77vBtug0ZQRUukacpcvvfzprnkrqZVRh0B3oVkWcjzZveWJnz+JDtUHpw1lqemx+c2j+/\nYCN/e21Fh/fsqNNuTwN0BgXdyK7IL5gW3M7NMq1Yf6eczMN6x+ai7lbkTifPlTMjoMh3MFXlvHW1\n/OTRue2r31xE7nQ8oxNKNMflHf9C/wcmwMd3B37rSJE76RhiITXnMSGb5DNIwJlNyOwNK2W3k+rm\nBN++YxZXPZfdCe+gKd45u65TDw1tKSsCTLfXIXSiDjuCGyPeXhx5DwpXzSiKO8ML1qUn8jrX0Sbf\nOJPhV3ez499GryLyroo8R5HrpuSXT+Z2tmnttLC/v26Rfa4j0s0j17+6jMZ4irmVtcz+vCbrOc61\n3lmx1TvX16HTF/NENNVVuK4DU89tG+9OMbyt0VP9HXVIf2fcUUV+8WOf8vqSLWxtascHkMo+I/EU\nVMcklC4OGlpTvLxwkzsIas0b4bUrA8d0SORJ611G2yPyXKYV10aefbB0w/R85NGdK55/9dRn/Pw/\n80mYTh6JHXfweQuC2okj70GmlQzBmEMYeCHFnSv7zly12rs2lujiy3amr+mYbzsbHeQKi0roBh/P\nX8ixSiWfMinrMf5O/cqiTdz77uc0x3Uet6e8lTeclHGOc7tP7WXJEFTk6QQQ0TzTioJ9nO41Ct0w\n0XzP6pSoq8v575i5igl79OGwYeXWdaTk27e/y8dR+7oddMiuZHPsCAVhqw7asql/NWzZx3MNZk5B\nO0FC6cX81VMLeGv51uwH2+isIm+PyJ0azDSt2O8xh2nFaSf+wTRlSMJa94SbOM8mOzDxdAXO6+hM\n+OGuxhc1rdwyI22W79SFESTinpQ5s1cReVGka8XNRfzfvSuY1SxX5IKU8EzkjwwStUzg6Q7v4bTT\nxg6ms9mK5W8U0VBwAIqEFN/KN/s4X6OK6yZFfiLvxAKMbHDMVM7go5sSTXjl6sjM1J3OTocEs5px\n1IhF5LlMKw75GB0TeboSXFfrV/nZn6Gjeoh3wrTiXCOnaSVH2b3Vs97bbUsZ7uxze5H+RCbdFz3j\nKNb2Ng7vSJF/+kUde/YtoF9RpMv3b0saGFJm8oeR8vwpNv734U9YtbU5eJyryIPl70kJ13qVaeWK\nE0Zwxw8O6PTxnSWTKcpc5HUDAqsBh175Mne9vZpBwlodGSN7I/TfI6JZnTDRwUrGbMVyOnRLQmfl\nlmBDCppW7JN9NtRc96ttSVLdTs4VvwrKNj1MGaY3AyDL4pX063VA5IffOJPv3PlBu9dw4Cjy7ERu\nd75cppUuxP2mP7d/8FPJ/rxGB+TmEHl7zk7HP6JnmFbaj7jJpshzvX+5AwOr6arQ7Q/dda9ll+Pe\ndzsOKsiF0+76sNNtJx2H3ziT0X98PfOHLO0ng8Qhp2klpfeMWQT0MiKHjnMxJHRvSXh7sdl+/FJ7\nBqG3YVZbHnqH4G6fuZqktBp0jOzTeH/7i9hKOt7BasFspOl00J/9+9OM3yKa4p6jZlGJ6fdzolae\nnlfFxOtm5CxHMGomC5HrMkBmzjG5Ft+1F0cOUFXXxmfr6zO+z4aYTeRN2ULn1LBdoDi6YfLb6QtZ\nV2MNwlc/v4h/vL7U+r0zNvJ2CCQXkXekyB3TSnvHOYNBRtip0j6BOlEPfjNFWw4ib/ENgrmOcZFW\nVMOmhhmLg1ExldUtPDW3ayscOzOGdMa0UlWXO+Ry+ebGwOzmwVlr+c3Tll+sJlc8fzKHEPBBSuk1\neD0oivKKfAfQkd138g0zGXHNa/ZqyGBF17dmf6FxLIW3YZudm+SJH3CCMsf+zSKNgpxE7lPktnmj\nY0WehchtkpxbWZfxW0hV3OgYJQu5pOcySb98S0Ln+leXZRznJ9tsHT2ZpsidjpJrxWi2lZ3bi1jI\nUkH/+/DczJmA5hB5K59VNfDk3PX8yl45+tjHX7B2q728fTts5P7mla2uoX31mNRNdyGQv/2ZpuTV\nRZtcwnJIPqPeHSWcIyopmyKPp7KXs9ln4svqa2gHjmnlsQ+C27Od/M9ZXDHdzlWSarPSCbSXLoEc\nJs40ddtee+nImVjXkuSEW9/nt9MXWo3wozu5/aU5TJ/X/iK02oaORUXKkDlNK05/sFYv71p13uuI\nvL0QpsUbGqi2436PvumdDJU5/to3mZMlkVSbtOxuim6N+MrKV7gnfCsACZvIC0V2NfD8/A3cajtH\nnP6doZDTXnI2HnA6aNIwmSBW0g8vmdK2pgSz11oRMCILuSRydGQH97y7hnvf/dx1wDrwK8ZsHT1l\nmAFV6qxCzBhL174Hnz7SrTZyx7QCQWUJ+BR5m2u+8MfnqnTeRt6eAz23aSX3OWfe+xHv2wmx/PX7\nxCfrufjxT101a7g2cv+syKQx4fhAchC5M5j6XkJLUs9q9vLPZjoicpkmyXWs+o8QrMNmN6++hGbb\nKfzWn93fP9/WnNHHslZxmlnDMLO3beu39tuSkyL62fkbYP1seP0qbgjd3+45AOs2d5zRM2WYPmdn\ndiJvTRrc/OZKd1/eXYFeSOReAx5ENfuLde7nk/85K3BsthVt2aaFrVhELvR4hnMnLi3SSLY2cduM\nzM1jH5/9Bbfa37v5RdpRvpCdPJxGkTRMno1M5fnwH9zfPvq8hkc+sp4zm/ssbpOYaUqe/bQqY+bh\nEEqGIveRSDZFnk7krmklnckf/ja8eGmHppWuwE/kGSSkeorceSa/rdkts0+RG6bktcWbWLG5ibXV\nni9ke4i8PZOJ33TkNxdssVMpbLQXeznXeH/VNoZe+TKbGtr422vLeW+NTYK5iFzPHEzP+9cchv3+\n1YxjmxLe87emuuaA1+3wwwjZzVOGKbPGVx/zj3f53r0fpV07m1MoOMM1zGBb8xeoI1NWwDxlO8BL\n6NhsouTysfiLaZi5nZ2++/5z5mpG/iGLHf5LQq8jcr8D6cPoz3k18rucxy6qykwRmu07x3yi6i0Z\ndtW4q8jjmWFJaTBzEGZ6Q8y2SMOf5hZgdyX7dnjZVMtpdhTOyq1N/Oqpz7j//bWB3928HmkzFH+e\nD4csI77oh0xnp+NwtTrnv2atDThT61u9uttRIveHjmYoHYfI9bi7itJvr3Qje3zv8vn5G7josU85\n/tb3OPqmd9zv/cVMJxz/s/sHx84+m/+9O4Ofcw/nGk4UyqxV1Xy2vsGLFsmxECehmxyhLKSUZve9\nZmtPmxvibrsA+OMLS/jTf5d0qtwASZvIwyJ7OXRTek7lDkIUs9ZWhiIP+mP89uiD/uL5eZZtauSO\nmUFBFfQzyMA92zV5pFpJGSavLtqU87ikbvr8Fl5dNCd0lm3qOVvB9Toi70podENbipAqOHX8IPe7\njVlyVDimlW3V1fz1v8FcxY79vDCHjdyP9M6Z/n275zqLbXKoQAdKjpA4yG0rdabh6QmagjZyiwz8\nRJ7McHbafwtYsrGRa19ayrSHH3B/v8KX57krRL6hvo2G1iBh+JVyRuSKT5E7U/2sirwD08qctbXc\n8KqXD8UwZcD+73/28de+6f5946vLGX51pgJOh78OnHHJ+S79XdS3ptBU4ToZ02OWHeiJVh4N38BN\niWvdVLjZ7vdJZdC8MXttLQ99UJlxPdOULKpqyGhVSWkTOdnr0JTSI7Z2UiFsamhzZ5PBBwn2J1Om\ntX3f7/6B6rt3fcBNb6wMmJICRG5HczkDYnpfDCDVyp1vr+bixz9lxrLsawcSfiL3DS7n/2tO1lXl\nHUV27Sz0OiJvb1FBOtpSBqoiAotlsqmXNlt1vzRnOU/MDoZIJXyKvCP4N1Pwo6OpIXgNIJSj4zjI\nNd33XyMdTkNPL0fQRm4dE9Y8k0amacVxdnq20t9suyrrPbvi7Jx8w0yOvfkdr1yGGSDmTIeg/T5T\nbbQkDPccB5prI/dUdLbFOfel5dhJL3OuQXXuurr2CcKGU7+GKTM2E0kf6P7yyjI+XFPjU+Q5wl2T\nVjvc31yVkbu7xTdzaW8xkp8Dne4AACAASURBVB/3vLeGb98xK2ORXMJwiLwd04pTxnYU+ZtLs6d/\nzqbINX9bS1qCKz3O3hEr/hlY0vDdXzqKXNjHB8sWWMWdbHMjYRZtaKCmOZERMtqS1InbdeF/J3PX\nZQYlgGev/7LxFSHy7KQRTxqEFKXdJfgAhm15LhJthAi+iLi0FLkTtdJemJTzWzrxdEadOp1eo/2G\nkMshBLnDoRxHYHo5skWtpJtWVF95DDuSQhEia7ZGP/EZptml1XrVzUkm3zCT1qTOsN+/ypM+X0aG\njdwhjpTftOIzY2RZ/Rro7N6FAp8MUwZNNB3MjrJh3JDSwPUWrK9nn6tecR2AzkwjV2isKR3SyE6g\nKV80Szrp+KNUQj61fsAeZTnL62w1l/6qkkb7NnLTxCM206Apnso6S8mZ1TMt9M+QMtB+jrnhVarq\nWnPuLuV38PsH1RWbLXOHQ+T+vmiYMjBYC99gcvtbqzjoLzMC9Qbwl5eX8cpS2ynaiZj6Fju19Je9\nUrUXEnnmd9Eci3VaUwaqKjIa/PfUt5mseMmGHLIqpjWDSNuw1qc7zpN2Y4NdRe41rJcWbuTlDvbO\nDJNiSKsV+5yuyIeIrUR8z6eI3PfPWFxiw+lMn29rDnQs/7M4duhISLF2p6n8gKRhBuLWzYTlJNzc\nEGfW6kyPv7+chtn1EMQN9W2s2pK5ICPdtBJP2vfxmVb8phiXgH0dNRuhpL/Ke95ZE3CEdkTk2RdR\nSY7bvz/FEQ3dkG4+nXdXWj4Pw6fSs0FxV+7mUuTe4JR+jRafc9P/vP2Lc6+GlBKOUBYySlQGvneC\nZ3Iqcik9J6dpLWLzE6qj8HPGWiet99yWNHj0o8oMG3kRcWYu35oz/t1/L/+z3viq1Y9Mm8j9bSee\nMgJOXaG3Bj6bMrNNvL+q2luE1wki//NLSzn4r29x97tf7sYgvY/IszB5AdntiW1JAy2LIv9b6H4e\nD1+PZpOmM6UrFm2ERJqatquoQljRCLk64GVPzOdn/87cbu5n/57PNc+3v03aX7QHubLqp9CwITAj\nEJjMilzGnaHbfN/JwO8OdMPMueGB0+hnLNvKJY/P857Nd3zcVeQqPHUuTDsRM97kEQu4K1+bErpr\nljB8Tcg/oCZ1M2ddTfrrWwy98mXOeTAzj7OfjBy0pUVcrNpsO6xTbe7xfoeoF7WScqNFsplC0iNW\nbp8ZTNkaePYss75sqjppmIQ1BVUVGKbpziacJpjLzJVRdkd5p9pgq2fH11NeWz845aUT/q7yPq2N\nXpI2//MWhHOntjCl5NHwDbwcuSogIhwCDYvspj7dNAOKPH2i7KTBSOmSErKslrS3srvxteVc88KS\nDCIvppVFVQ3Ek+nvwKq3ZA4id4SYQ+Rt7SyKEllSPGRT0mqWGV4uvLJoMwCvL9nc4bHdid5H5FlM\nKwUiewUndBMtzUbux1Nha7snzVXkbS65O3DU+m42kadykOXzC7Z/8+dDVXsloh4PzAgcUj9O9QYI\n//TT74hK6GbO6bp/ejrPZ9vzk4njO4iFFCsWF8seG1ClycwO2aiVu3/7p+FtKSMnkW+2ydWJt/aj\noc27hpMbw1FVumHyjb+/7dlS9TY3xjyeMl07ub/MT738Gmxe3ClFng7/dbLN+rINnCnDJKRa4kE3\npWu3dtqt8znXMn//sxmmRL50Gdx1CLRZ723RF16dnaBY6Y6Hiy+4JXw3g967wv3N/879uXtueXMl\nd76dPcf4s76Q12TSeg+5TCtVdW18ts6OrJJGTt9V34bFLIz+mBOVj4M/JCwi92e39LftItFGfVsq\nQL6V0bP5o/aIdbpv3UAyi38km43cWeXsmFLNREvGzlXpA2xYVbwAg04QuXudHUzj3FX0QiLP/C7q\nU+QRkoHPmipy2sgnKFaD1mzVUSwyTStOZ95NWB1pR/NspyNCkiHCscGlCPkUUDZ7ueLLseI3ZZx2\n14f85NF5GcdDUJ35G6q/sdXZ4XWFEc01SZjJtiCRZ4m73Rzbx/07KrzytCX1Tjl507HNF85YHNXs\na1n1UN2cZF1Nq9fhU20BgnYcTX67/qUrzoN7JpNMZRJSNtOIEDBiQLF9He/a2eKSsypy3SSsKqiK\nwDAlrYngkvpW1zlrnVtBHVO1aa4adp9t2X8ZcdWLbFtuhxDWVZIyTBZUemGpjpnNmZGGmm0xMfs+\nRi3+m3tcxOfAvu2tVW5qZqsOvLKPUSrdv506jOQwW55214fc8po905S5VzaWN1m7bx2upOVNt4nc\nX4f+9l5EG81x3SVyR2BdoFmx2rlMK049ZrORpz/bwrUbOwwuCGtKoL11Fks3NXLdS0s7ffyOotcR\nebYl+n7TykeRn7E86u35d1nqQX43ZxLPh6+xzs9i93RNK7SmOTul27j620R++8zMRUGdxUhRyeOh\nv3Cx+qL73S+16d7d9LZAw/LPDs5TX2eI2BYIP/SrpRXt7Bvon14GcoabfiLPJDoj1RZQSUoWRS4C\nA0vnFLkf6QSwtdF7l64iT+jw4T8xt1oE5BDsRyuqeGXRJmLEmRH+DSU3lHOk8llW23Z5bVoOm4dO\n5Cc1N2Yc96PJe7l/++u6RLRkHJstSihlmIQ0hZaEwROfrGepHWvsqEZPkVvXvjr0OOdrb3CMMj/w\nbLRsYzdRx+dthdbnTZ8RTxmB9tGfOi7XnmAPsSVwTV69nJGVlnJ99Lv9Gaxn7iL0g/sthazI7Irb\nafdFOWa7/mMgt1PTcOsozTHbYCejS9YwWnzOe+FfUCa89lUs2vjo8xrm2mGUBynBTWBak15OJf9O\nWY4pSGaxkVul8Mp5hPyEVdFzGSOC0UuBZ1SF+06aW7oWN/7ArLUdH9RN6BYiF0KcIIRYIYRYLYS4\nsuMzth/ZMsr5TSt9RZBsztBfAmC8soawpmSdKjqEWURboHEWkHCJbDfq0NB56IPK7c6r8ErkKiar\nS/htyMtN4Sh9gFRbE0OFZ1vzDyp/Cj3MrMgv2M/0nCgRu9GqGO0qC/9Sbb8Dst+CO/mh+iZREtQ2\nO/lBfM7NVCJAiiLlkJl0Vc22Bo/g/OaH5rjOtA86bsjpqnZbYysHihX0p46wphANKYTrVsEbV1M2\n83L3eQFGi7XoqQQTlZUMUyw1eob6XiD1roNBdXODX6z7gMNb3+K58B84SfmYFZFz+bt2T2D25n/2\nUrIQeZaBKmErcscJu3yzNcAW08pNoXv4Vu1joCfdcz3lmwp8BiijiRppzQ747y9IJOKuOS0pVcYo\na/mp9iK3he8CoF/DYmj0HOthUhzx6nFcuPD7GeX8cI1lT48Y2Vc3Os/eXtitv83lcmo6JiQrQt87\n5rmPl8Onj3J71fd4KXI1eyjbAmRdRBsltHDDy4sAyX/Cfwlc95dPLmDENa+xqaGN297yxJVTP46N\nPN3n4n+newrLEf1DdQYg2Y3M9B2JlOkO6Bu3Zf7eEeIpg988/dlO3yR9h/ORCyFU4E5gClAFfCKE\neFFKuVPmFU4/GyK8Kebl2pN8Zu7Dtfq57Z5bFNGy2rk8Z2droHGW0uKSgiZMhohtVMqBgRWMO4ov\n5G7u3+pLv+DBsKcOsi3GODX5svu3M+19JHQDByvL2TfxaNZ7+POjJ3QrLFBRBEPm/Y3rQnBd6CHG\nLbkPKAo4AGUqaCM/ctmfifAPbgndxYnqHIbGHw/E/k5QVlEs25hljuGhDyuJxbcxEJNN9Mv5/H47\n8yRlKTcu+QFEYJE5lKuVO9mtJMpuW+3pdLS/VU8+5/SvtaeYZ+7nXqNWFmeN/z58wwP8rxqnkDgT\nFK/jH6Cs5s7w7QCcqb3H3xRBX2Mb/6vO5GNzpHtciWilXDYwTNngfp9NhaZsZ2c6TlVncYb6HrS8\nB/NGY5jjrePtLuisoPTXd5loCUQNbdnkmQI2ynKGKllitG8e4f75ctiL8VcxXNuwg4a2FDEzc4AC\nT21nSxY3VGzincivmW1693LqooQWGikABFJKTHtR1oHKyoAwOUebQes7aynwXbdJep+GK+uZrf2U\npXJPfpS8PHD/QtrYYGdC+GB1cBeuUBqRpyvybLO1s7R3GKZs4EBlFRtlX45L3ESrHa3WljJQQjYH\nGHFYPQNpmpTQTCNFgBVy+lmWFeMAby3byvR5VR1nn9xBdIciPxhYLaX8XEqZBJ4AvtMN180Kx9Y4\nK/IL97sDlVX8SOt4L7xYSCWWJcLFabR9RTPnat5GtWWiGQXTjSW/KXQvE8RKtjXF2a04nDM0KzuC\n6s2ZBTi5XADUuuAU77rQvzKuUia9BnOSMpt9RRWT1SWEhJFzAYtfkUtpLX449Y73A8cMtPOuf+LL\nvjh39abANQtSNdwf+gcnqla0xH6iCkV49fPH0KM8Fr6eR0LXs0diFXOiP+Wj6KWcr+Z+N/6czhOF\np8gqRAOKItitJEpxcyUA87fZq199IZgHKKv5a+hB93OdLM4Zi39N6DF+FZrOUWr72/79seWvXBN6\nLCAWHgr/nbnRi3kifJ0bqpdNkTs28nQk8G1gsGGeO4ClpEXkjuJXMWm0Ca0PTZQoHpH+5uG30Oyo\nqlqKcz6Dg32VDe7f/mdxsHJLExEzhyK3BUypmoAnzoalL9jlM3gn8msADlG8aJqkbgmdhdELOU+1\n+lDKkIR0a4Y8QlnP99R3AvcoaAyG6Km+iLHT1FnERJIDlVV8Fv1x4Dh/fqUP08JgnT4ps0StQO5F\nXgfag/sgUcucyCUB30DMlrt7N82Fx05H/PtMFkZ/zAnKHPrQSFSxUlmcM2lP95zdSiJMCX1GyC6P\nVT9bmRa6Eaqy+7J2BN1B5IMBfyaqKvu7AIQQPxZCzBVCzN22LXsekc6g/YWd7Zs8iqMasTSbnxDB\nKeKpqpefolS0oGGwSlqPM1FZybORqex3z+68kTqfldHzuCN0O99RZrG32EgF2Vd7QWaI5EBhKYn2\nBoNj1WA4Y4uMBPKi/yo0nTcjXqTCCcocrtfu55HQ9VyiPs+/Qn+jgnpbkXuhWze/uZK1VZYpYoZh\nbdThEEnI9K6/vGprBikeqXpOq4nKSjQMmollHPNy5Pfu56mhR/ix+l8uVZ9llHDMLZIhYmsgCqi/\n8BJOLTP3QFMEA0qiFCcsc1OdnXbUX6ZDlOVUCG9wa6DQ7awpO5e8IdttNAEoimA3aU25i8ju3Ho5\nchVlNFH63//lxuuvZu7ytbDsJczZ92FKK0/M4cqigJmkjz8Er/bzjKiGvsKyvyqYrjmlTDRTSCto\nMfcYZ5bWLIN13hH2tO3oA6nh0cjfKKOJcx6cjaan+z2CC9PKZR0sfwmeOhcpZc41G6lUkqfCfwLg\nYu1FFEzLX6B7in+AaN80kStCJh33hG/lQLGCH6mvuhFQDkJpUSu5FPkic2jO6xeJOB9HfkYJzQyg\nJqupzinH/OhFPLn127wXuYwBehXvnlvBQxccxMV7V3O/eiMjllqhwynDZIRYb4mIHBtr7wi+NGen\nlPI+KeVEKeXEioqK7b5Oe0v0bw3d6b8j6cReEg1lEKoqLGdGvbScSluktwquFEuRV8tSzkpcw/cS\n1wR+AzhZ/ZjbwncxM/IbPon+lLFiDULAAWIVh4hl7vFlabG0+wkr7CmUI043G1bI3TO+a+x/kPv3\nXeHb+R/tbY5UF3FF6CmOURcwSVlKUjd5OXwVj4X+Clir+frZxLFWDrSex/Yt+O2UEVJZFYxDNP1F\nHSoGrbLj7beuCv2HX4emM63USi96nvoGsyKXITd7A0N/Uc9KczBzzOHWvYVgQGmUcsMiIefdtbdQ\nR8GKR25TCnjDPBCAx4zjWGZm1t0qbXjgc5OMoSmCsLTIqq8SdCAPj09z/14Q/Ql9173KbxP/ZMIT\nE+DJs1FetUwA+zR+zGPh67lQ9cxgfUUjcRniJTkZWra5ar5YWIq4L9a9NGFSSwlgtZkCGccoGwpA\nPxpd0dFiT/07C4ckfxf6N0eIBRyrzCeeMtm4OZhjJEYCITwiH4z3e0NbKifZRutWuLtpDRB1XKK+\nQFI3KWn2Zpl10jJFPKB/q90y3qN/2/3uNeOgwDGtMkK5aOSZyJ/4Q+hRmpq9flVOA5eHngI8In/6\nrQ/4k/YQ+whrduK0ndfTrpuOPqKZhdEf83H0Ukp8/pF4xdiMYzdouzNEVPPTxWex51NTOHp4f/rq\ndpttsTRuyjAZZpdBlu/b7r23B91B5BsAfy8ZYn+3U6AqIkAu/9aPBiznj19NR8hsdEVRLeAZB0uB\nacJgqbknVbI8YJcuEy1omBgozJb7M0fuz5GJW7hfP5E79VP4t360q/ocDBHb6BNVeS7yR56M/NmN\nKOhj3/dXyYsA2F9YkQTtOSkbZEHgc40syThmzclPc3jiVj4wRjHTGM86sz9tPnNNGxEiJBmlrONw\ndQklUYWalqS7cUalHABYsw+A4cKbXEVIuQ3/p8mfu98fmLiXGllMOQ2omKyWg5lpWDbfFeYQVil7\nsUWW0SALWG8GB+1wQQmnjBvkxkCbGxYAUEE9J6ifUEsJzTJGkWhDUwVlMY2BWLMXh8gVTB7Vj3Ov\nuVjdn2tS5wPW1F/FxESh2a6/NqK8ahySUXd3m9/hNv277udVcjCqohC2VWcfu04+Mffj1MS1JAiz\nf/xfrDEH8p4xhnv0k/lp8ud8NOg8wJoxAVS0WmGtewvL8VhGEz/RXiYqUmw2SpDN21yncrEd1ug4\nvRVMYrFCGmUB5aKBQtHG57oVq99PeETe2gkiX6HsDRdYZq0QOiW0cIpqpZjt388SLMVGUCVPUFax\ne5+CrOaprU2JnIp8fmVwlj1aqaRm/TIO1b1FS7rtD6j1teNTE9fylj0rdPrrVp+YmpoK+r1OTV4b\naFNuyCVwourFqjvlvyV0J+dpb/JW5HKOUBa67blFeH3rgPg9/Cj5G25Jnc4Cc++MZyuWHmdM3Xgw\nnxw5zf18beocpg66h6XmnsFzdOt9mvZsSk02cWXoCeplIUY4sx/vKLpj8+VPgH2FEHthEfj3gR90\nw3WzQhFBe58AHtOP5QT1E8rxwoOyOWmKoxq3hYMhZ6oQaBgkCJGQocA0vZRmVMyAk+gLuRt/0X/o\nfr5KvxDwHEAhdE7otw1s0914sYYv5G7cYa/OrJIVrDcrGKZsACN3djmAJgoo9cUv18lMu6hhSqpk\nf85O/T7w/XDxBa9HriSEzmjhRY+cNqSZaasLuMJWLmttIndURywtJt+xlW6WfQAv90yNLKGfaETD\nREflx6lfs7u+hfWyP6PKS1my0XsXAsvz/3DoBsbY+2064Xxiy2JgIN9XZwIwmGrmM4yhbEYRgiLN\ndKOSHLOYat/Twd/7XcfHXzTx59A0QjaR61Jx88wnCNFAYUbdVccFz5pncpf+Hf4TvZG4tNYcOA7c\n7yiWH+Fm/UwWyGGANSgcm/xH4DoNSjmRAS0M22Qp8IoWKyueIxoO9tmSq2UpItWCath5s21F7rRp\nFRNN01hiDmWiuooSWqlUKxiGoEw0u4N5s8xN5AmpERE6KRGFIstBHELnIF85fjShjLtf9wSFgzPU\n9xj2g4sZ/ESI9AWZKzc3EsmR1nbpmnXg6QfaCPPoi6/xJ98xziDQ5DPFLZD7cJt+Gseq84nY6xD8\nRL6VPoH7VMtSjkzewkSxkqcj11LcVgW2M/3bqpcHXcOgiFbGCc8Ov6fYwnL2AEAKr/3UUcJMcwIz\nmcBtxumAJQhOUT7klvDdAXt5jSxhrjKWM+P/dr87WoQ5MXk9zw97hfFbngWgT9IaxAtrFqFyOnsk\nrDaxxBzKQTthrdAOK3IppQ78DHgdWAY8JaXsfOLjLkIIwVnq2+7nmEiQQiOEji69xykgkWFGcRaY\n+JEyTDQMdFSSaIGGakUNGIFl6LngOK3CQueQcm8QGWKHOO1lRxgsl3vQSAGFdtnC6FmnyQkZctWa\ng3QHV0qEcy6/dqIhQujspXghjePlssBxn5r7YkpBiWilJKoFFvVERcoNvdpCX2YZo/hx6lcA1MhS\n+olGVLvuANbL3QBhLSryQaJgoJJCQ7VzhDv28NgmS7E597lc/wlNssBanKUIShTvHfpNKyYKzxiH\nW58LSt3nVTHQ7HeWtL9LyBCPG8e613lUP45Lkz/jPdOaJicIowuVkNDp1+qtehxs+zFMGXz/Q/oE\n7dOzVlfzaVWLq5ZLWy1n3AHKKhRMl8CeCX2bGttsUmIrYcc3YS0Ks8I6VS3EHDmCUaKSIhFnZXgk\nqGEi6G6cdAu5beROXRhazN0UISQM+grPVBQzbCekWB84N4TOqEEllEUz23zFovtzmlbKCJqh4jKM\naLDMh1ekLLHjnOsXRiMGlLifo/bv/kHXROFXyYtcm/YdFxyJRGG9tFT5t9WP2K9fmGd/uDcHKV5a\nWQ2DvcUmwsLgXWOs+507m7frxTGppsNApaK02C631yeqZSmvpOVOct0dWsRNW9Cv1TIpFTau4TT1\nfYpshf4H/fx2NzPZXnSLjVxK+YqUcj8p5T5Syr90fMb2QxGCQcILOSoggY5KGB2BZI1p2Xx3E3UB\nx6YhRYD0Pi36Bo2yAN2UhOzQrFTaBKXMVeQdV5NDGmF0Ttzbu87uttKqk0U8ok+hkUIShNzGEUKn\nhYKM6zUSo1QEiTzdwdWmFjNyUAnHjOifszz7lUcYIrZhSoGM9WG/xBK3Md+qn0acCE3EKKWFYf2L\niPkabYSkOxVNSo0fpn7Pe+Y4AGoooR+NWeunIkeSphQaikwhpEkfu+MX1S2ljCbKRDONsoCPzZE0\nEaMYaxs3/4KUmI/IDRR+nbqYveOPEVKFG26m2dE7SVP4CEOiozEzbJnhWojyX/Mw/ItUUoQIYbB7\nbdpScsDwHffpNVP42xmZdtIUqkvkxa3rqZYlVIhGDhQrXWU+o/wcV1EXGdbMzzG5FYgEU4/djX36\nxQhpGlXSS33wkXYIUrHESsh1duZW5I6irSrY383dHkKnn2/GGrXvP1AEw/eO3LvUymeUZb9Ttb4y\np2nFP0iAtSHLnmo1aDHeMiZY97RFgu5rL4fu04+Uu62cvZYBhUpzN14stNTxs+aRnJa8llVnz+HQ\n/ay9BbbQh3eNsZyhvsfp8nUGR4JhlBrewHW7bT4rE82ef0UojI4/wGGJf2Z9HoDdK6xMlo7PBCwx\ntWhDMNTQ2XQkEola9WaaVDR7s59C4pTYRF4tS3sukX+ZUIUIkM0ScygGKjGRRBWSjdKaZj0bmepu\nBrG5aBSqkBQoPgIrHOxGjKgYpFBJ+kLE4jJEiWjpApFb54bQCSUstbXU3JPdxVYEJqW0UGfHncZl\n2FX+IaGTEuGM6/ljai9O/oKzEtcEzAkAGwusON6jhmc6j50ZwvCKCIOpZitlUD6CPka166itt51P\nTRQwaVCIkliIKEka7QHD7+w00+qgXha6HSO9fvbomzkwWXVkKfKY2YQmTGabIxBIBopa+ogm4iGr\n4zTLGFGRQkOnyN4rtVHGqIg4C6BMuzyW0cbKpSNISRUNA01YpheHINxIBjXiliMdhk3EUb3Ry3du\nw//sfQvDgSXvbn2jERYGZTQRTjXyhjERgKMHJt2IlZI+5bT5thXU0CkRrWwtsMw254+JURxWUFSV\nRp9SrEuFMJUQmo/I27ORf2yO5PvJq3ml73lgm7LOO3gwfUUTbTKMUdAfLdFANKRk+IxKQk7CMZ0t\nmrchS40splWXrqreJksD56Vfp40Iuyu1ULa7K5CcQcBEoUYWs0YOoiCsuu3aMTMaUuWo5C081ecn\ngfqlZBBCCNZefyKfX38y/2deiSkFg+RWdlMtIm+tsISGhuGKhRMnjSGhFVFKi5cITVFppqDdeozE\nnH4QVOTp2Gxv31dSZLf7urVE9UaetWeMAkkoXk1SqjRQuMO7Z2VDryNyIawGscDch+MTN/BP47tu\nhwWLlBw4ttUWzVJBpbYiaZIxIpGY3XCkZ1qRXgevptSnyDtO0u83ZdBSg6EVUCXL6SeaKKYVRUga\nbOJMEHIbdQgdXYQyrqcVenbCLbIPs+X+gee8zTyLF/exLJDZ4pad8hSoksGimg2yHKIlxMwWymz7\ntBNFEJdhQjKJlJZqarSntn5np5G2xNoyZxlowqAgGmHMYK+B/2jyXhw+rJx0pNBQZYoi3TKrrDUt\n+3wxrZTRghHpw6DSKHHF6kBhvZUCe2XhNllGlAT7VBTa5hzvmUPOVnaoqFgxvaYUGLYj2olPVjWr\nnpMys751NDQMi8hjfaCv5/S68BvBKAN/zmonIZUzcO5hm9KWScsWe8mhFe5so39ZEQn73o3NLZTZ\nZpX6mO0oa6u1ElCpmr2oxkJTQrcVueEOSgFz3LkvwE/ep7qfNXgkCPGxOdJqtzaRa0Knn2ighhJk\ntAza6imLKJTQynTjSP5rTLIiMnwZDRcrVlTPS8YkkoSQqTbXju13foMXdeNAYjlnKezvDpxOmzek\nwsGJu5iS+Bth1RtwHXHjtLX0HZCcVd1CCIQQlETDrJUDKDdrrLoD6o+7mQ+NkWjCcGc7oeIKkloJ\npbYwsy7SMfVFI14df6IdyPj4vVlNWlubLJ4pKbQH30YntNeaicRIUGLUU0MpIDpM1rY96HVEriiC\nmEjQJiOskHtgoqD7Ikf84UqO/bFR7Wt9Ni0CaaaAcCSKIiwS92zkXgevliWWjVwYGDJ7NflX8AWI\nvLWaVKQvtbKYPqLJbVCOPS5O2FU2EZvIP5H7s3DEL93rDajon3Ft3ackXzEOQUZsG14oN5HHVINy\n0cA2WYaIlhLWm9xQyAZnhkCYkExg2nHCrTJCQmr8XHuWIs1qdZFwiJvOHOdeP4lGWOgMKY3wzdGD\n+fa4ge5v5UVhHvu/zCiRlNRQzBTFhvUe1jmOVtFKmWgiFSnlw98dy1GjLBJEj1MoLSKvphRNbyUk\nrBA9/yzAyW6po1qDi/0+9xtkOcoGFmmBeko3oQHowjJdRPQGi8hPvMn9bUBZ0I7q35ln+IAS+5pW\nG3SiT9Y5K3bjjZSJg1x0WgAAIABJREFUZuplIX0Kwu7CoAgp+thT/8ZCh8jrwNRRVI0GnyJvTqQw\nRSjNtOIjlFgfGDiW+AAr3FKELAIqKwiBYs8UpU4/mqiVxZiF/aFxA4OiSRQhWWwO5dLUz9EKSn3b\nt+mM2qM/B8bv5tepi0jIEGYq7pJxMzEOjt/JrfppAO6zOAijWyq9oI+ruJ1BIBqJEA2HMVEIawq6\nPQhGfIo9vZ4huI8rQEksxBbZh36yBlotIg8Vl7vtoI9oQpcK0eI+JEKllOIRuePsHDUodwRJLObV\nsSlU6n0+qhNGDbDLKBjW3+pHRQX28c2WP2yrLCMlVQpEwhpEbbPazth0otcRuSoEUZLuXppAQKku\nk3vysD6FOllEuR2B0hy1pojuZxlzp9l9o5ZdVZdBG3m19F58RWl2U4FfCTtlCAsdWqoxY/2oo5g+\nNLlT6zq7IaQrckNonGP+kYVDvWRf7r6U4HYEP3klTOkqUTWtwU+/6FBXBcUUkz6i2VLf0RK0VLMb\nM+6YVuKE0UxLVURJ0kaEeeZ+qEJywUSLDD++6pucceAQXrr0cI4ftRspNGt1qqmDUAPx/dkSm1l1\npCHMFMWGRXZOxIyjyPWIda9Q1CIxM9lKzF6Us8zcA0XqDMbqJH4HpKOQLUVu2ciFohIJW3WYsDei\niISsOsluWrEUbyTVCNEyCHtEKpTgjMyvyI+z/RNO23Ginvbee5j1XaKRMpqttlAYChK5rWKbC+2B\nq7UWTEeRe/dvSRiYioYmvJw6t557uFcgu60MHmi18+8fOJjLjx/O7070bOQaOn1FI7WyBLP/KNiy\nlJGlVr047UANRbwUFkaSAX2K2X/YPiSwBqB4W6srQBKE2Eof3rMdia4i/58nAWuhW5lshIJ+bvt1\n2vz4Pfu6S9ZDqvD9bl1bUTVOnzAkY7ee9DxLJVGNzfSlj1HrKvKwTeQqBn1poo4iiqNhUuESykQz\n359oLe6rabXuv2RjI6cdMJgLj9iLdESiHpE7xH/RN/ZhwR+mcMPpY9wy/efCSbzw08kIze6zzdas\nrJ4i2ohQSJwy0eLOgPM2ciwV7JCNA7/pY4vsQzMxCmlzifvgcaMBmDzAqsBmYpg2kb988UGugkv4\nOniNLKXUtgGP3aMfb/zyyIyyBBuaIClVV5HLgn7UymIiQmeQnabWUVkBGzk6hgihG5KEbrpOLKXF\ny6PhEI9/oDFQXIWSTpsTh/b1FLmiU0Yz9RRBtBQt1eR2Os9mH0Iz45hSEiNJnDD/NQ8FoFy1I3Bs\nMhs9uJTzD9sLHdUK0zNSoGhZk5mlI4mGMJKuo8+JYS8WrfQRTRhhy5wUiVn1ZCRbiZgWkS8wLTvy\nMFkJeIMbeMpNR7EclmURBvctpnLgCTTIAhpGfJ/vTRziqi8dNcMcpQuNkNAJp2zTSsgbvIUaJHJ/\nfvsJe/bh2UsOc+vbyZI59XtHECootRV5Cw0ZijzpztRaioZaF2urA2mgqmpQkcd1TGGbVoSORBCJ\nFXkFslW3sAefKHF+evQwSmMh+70Jm8ibqKEYOWAs6G2c0s8yAdTbg4ZQI9b7lBISTRApcff+TBAi\njO62W2cv27j975hi20a+7zdJxvoTIUWJTeRO33CIWiiqa16IaEqGs3P6JYfzj++Ny9hHID0ddUks\nRI0socRsgNY6CBUSjcVcRV4mmqiXxcTCKv3778aeBUnOPsgicr8ouvms8fz+pJGkIxz2TCvOLKEw\nrFJWEHbrZcSAYiqKI4zbvcwTX81WlFi9LKaFKDES9KHJnQF3deeszqDXEXk0pBAj4W6YDMFO3UgB\nLTJGWBgMpJZmpYRIgdWB1TbLQ98kY2CPnuUxSykkCAVMK1spo0S0ESaFyEFU6VM9x25MSw2isNxV\n4M7CkHqy2MiFjqmE0E2JaUq+k/wziUMvQ0l4EQaxiDXo+GNfJYq7W1I2AWyiWJE6RgMhYVhqIFKC\nIg03r4pfkYfMhGsjj8swLU5UhFMOxRtEQqrw/Al6HBQt64rbg4YGY4CTaGCk3FCsddJSs31oplS0\nYkQdIreVS7LVzQWyUFo26z3M9e7zOdBUwZA+MXQ0VAyKlBThaIzR+49iXOIBho+eyN/OGOc6KQXS\nMjv4oAvLRh5O2aYVnyJX1KCC9w/gmiKYsEcfd7Dtj51moKAvREpsRd5EPUUURzUS9mKtiEi5M6NU\n4SDQopaqNE0UNUSzzwaeNEx0O8T2qKGFiFAMVF9kkFM+p8z+/TCFADWEKg360UiNLEUpHWK9nyIr\nosrx3aCGwEhYeedNHaKlFIRV+92F7Fz/Vrt1bP0OkRe32vHoioJUw3ZoqmkTudU3nHOF6tV9SFXc\n/rtPH/s5nJDJNEWevjtYSTREKxEiMg6tNVDQl7BqDQyWIm+mlmJiIRW1oC/laiuF9q2d9nPc/l7S\nuvQmHA57dexGRdl9PhpSeeRHBzPtgoO9E9SgIm+gkFYZoUAkKBUtrml1J/B4LyRyzYpQifuWhTsj\nuhkpJRpSmTRyKAB7Klto1vpYnQpg8TOA5RA1VGvaJBIWiTRQ6DqsqmWJS3Ilog0UlZCSWVXpCiGF\nZkXCtFajFlVQay/g2VuxiLzOR5xOow6TwlCsBqCbkjVyMKmjrkEfdbp73ZcuO5rKG05i8n4D3O9M\nKVxvucjQ5F55YglrNlBHMUQth+QeYiumFK5jOE4E1XBs5AnihD1vvr0lF75BRFMVb3aQsuon2xZ8\nT190GJcd5zkKU2hgJNm3KEG9LKSNKHEZYrA9Y5Exi/hjBVY9yVSciL1wZqPsB0KhVFqzCb/zVVME\nr192JAXRCCFhEEvWQGEFh+7Tj3lXH8dxI53OKtz/9ykIRgoZNlGGkzaRh7xpdaZpxWfW0YLOzgrR\ngBkqsGKKoyWuIq+TRShCBEwrbtqGWBnE+lqq0tRRNQ2JQkJqbBhpJYz6oiGFhsFeBW1QWO4KEcBV\n5O4sIpWW0VANE0o1EBNJKztk1GqXosEaFLWiflbkkxq2nJ1xO7wuWkrMUeQyRER4q6UdAm/Lkp4h\nEolySD/bRGMTuY7qqnl/fYY1ry0NLLTfqd3W0vtcOgGWxDSPB5o2QawMIaywU822kdfZihzbwYtp\nb+xhU98D5010r5cuzLSIf9av2Md47e7I/SroU+h7D84A1bwFXSsgheaZVpxZMZ3bjL2r6HVEriiO\njdyvyK2GoESKWP7nb3HkKMvetYfYQrNWBkOPhH7DYKO1uUCNLEGP2alVa6ysZ3Wy2FVVjnnGgVBU\nVLUzpoOQ5eDR46jFFe702IlkcOyeVhx5CpCEMTCFbTqxczorAsyjvBSkTkcdt6cXZmgiWFdrKS+/\nkvBvtJtEI9LmxLEXW8Ril6eRAleVWDbyuKXISdJG2IuKcBW535QhfGYeCYqK2n42MwD6lhSBqTOs\noIWivgOZuGcfmihgD8XO5xGzBtxYoa0QU61oyXoSMsQPDhsO4SJKpb2zjM/MlDIkhRENRbUUeUGq\nDgottd+vKHtMe3qsuy40oqQIZTGtpCty/wDumGg8G3k9MmoLh0gJxOsppYV6Mom8VLSgS4WKvn2t\nd5NoBL0N1XZWDk88wm6nWyuRHUWuxa1Byu9Dcf8uscMFS9Jy1ika4VarjmsoQbWJnHpLRT992YmW\nstTC1l6hfiIPe6YVf9oL5zmcxTtmpBQOtsMF1Qglur202SbypF+RB2Z3Vril9ZAJp8Ktek7rcxmm\nFVuRA9C4wW0/CRkiJpL0FU3UyiJrMIr1ATNlmYwImlYcOO/y0L2tMjsLgsDvgG2nnTvvoWkLqbAl\nSlqIUiHq0YTpisM8kdtwyMaBG4rmjIh2NMcgUUurWgKK4oaT6VKhjiJSUTs8bpuVJKpeFrqmlc2y\nr2dagIBZwY/015FEY4BtI9WKyokrFhnsLrbSKAtcW35chlGEJGxHIZg2UXtELlD8KtB+rqEVnofd\nRHEdNP629edTR/vqRUVttWztlrPTUuS7K1vd2YFVnhCqYZlWYrZpxU2E1bgRQoWB0cI/HXbqpyMT\n+c+PGcYPDrW2hRNNm9CK+1MQ0WiUBexuD3ShYqsDFdpErhoJRGsN4dL+XPPtkRAupL9hHVvrS1fg\n7t+ohKyl2XodFGVJzCYcRS45ZdygwE+Gf1VvumklXZFrQbMOwC9PGAXYRB6zQ0cjJVC/HkVI6mUR\nqiLcNhYhxWGDNMxoGZP3rYBwEcTroa0exVd2TVV49/Kj7AVHBlq8FgrK04jcbp+7Hwzf/w8cNzX4\n3GoYrWWzW29KzG5HtiJ32kVWRR4wraT4pjqXlFRd0dNKlDHxB6i+ZAWcaG8vp4XdyA1nNqwHiDyo\nyBdMPdH6oNvZJh1FbhPr3uWF3Pb98UH1i2Ujd3mgYYN7rxpK6EuTbdL6//bOPdqOqs7zn995n/tI\n7ivv5JLwTAjPPDAIKCGAGBBmBG3oVlFQVovT0raDLTL90O6Z9m3b6jjDUlp7je3Y7aN1bB0VBZ1e\ntjxsARVBwEZBRG5IbhKS3Jv72PNH7V2165w6555zz33VPb/PWlm5p06dqr2rdn3rt3/7t3+7O/Bn\nu3ty2M7WTRJye1//4ILjuefWXazqi0JqnfDXWv83vH4Azz/DRDH47RFTDCcwupeeulbArms5EXbp\nXnrKSv7schsW57qYRU+ksvaBt2/rfbIEQ4ajZfuwWCHfT1doVSVZ5EkhQ5U3ZMxko1VGOgcYywfn\nXiHDMeF0A0VFxmKuFX+VlZhP3r1IPN/iip4OzjshqMPm1UGj+e+/t4WXbI7cL2PkyNgHapgusI1r\nrewJB14gsMizE8FgZwcjHKEYWeTDv4LO+MIQgSB5LzdJdq1U4Rr6gd9AZz8d+SwH6Qinwh83GORe\n6+gIrluZUTi0B+kcCCJhCl0sm7CCRPRSGxmzq9mUi5zUPRpE03TWzrD5lgtP5Mqta2Pb/NBOyr2x\na52pGOx07gaIxOa4lUH7Ws4wYq1QSkvgYDCg+PqLtiISCMiYyVKUo3ROHqDQ5az3bhh+EjDkuuIx\n+IN9HYyZHDkZJ3t4T1C3nNejyHj+/o27Y24hIPCRW4t8n+kOXhoQiG1xafQicIOdzp1WWkqHN9jZ\nLYc5I/M4X5t8Af4Q+0E6KOR9w8Mrm/ORm2w4JmC8sheymSCaKJML46/dS9RZv6t6SlxxRlVmbJaU\nPNfK+JHwGQ+CDMYoyAR7nWvFuu2ckH/w6q386y0XxI7nLPJsRljeXYq1gdBHXtciz4fnGC8GL45D\nBNkaIRqT0sFOCBdAHSHPlsEe/vrqM+h08ZvuQtobCjBiJwO5t/Veggs8WeoDBIaCqbTDpitMtPWE\nWcHNL9saHkMyuarBMQBTYZOPeRY5HQNM5iOrcRhfyINjlThKh4wwboXcLbKQEYlbuK5e3gMrnv9w\nXV8HT7z7UnafugqfMXKItXIC10pkYQybaiEvTz7PEjnCb0wfh12PZHIssAAraj7mTZ4ik2vItRIJ\n+VPQuYyOQjZcRAFA7D3KFIL7ed0LVsKhoej8hU76JoIuu2+Ru7UbM9k8J5St3zlRyIMyLi0Hg9fr\n+6Nzj3tjAKH15qpXx0ceRr/Ye5SXCTIdnmvFsmb1mvDl7NwUpfGDkcAUu2A4yNGS645fbxGxkRjj\nyJHnghdrbLCzum3GyOaRwzZyik4rlFJdVzfYmeRaMXn6bbTTDyer07DGZrv6/vuOfj593VmMkWOZ\nFbT9XdGC3eFcjEw+6A2sOAWWBO3YWb+VvmtHzCKHyCL3Zl8OY10rdiDd9RQGlnSyamn8hZfPVUxE\n8no9k81Y5MBkKbivfvTRsIYfehwMBg6PUGLXphVBA8rER7vpjB6EkVzcIj8oLtQqF+y3J0i0M0xn\nmC/452YtK/yc6Zks3aU8P3vXJbGiVFnk5Ci7pFOd/cGgl8W/oc6/v1aGWCYH2FMKXCQj427F9YpY\nbCfgnqAs6556YQEXWTJpJHiAS5Gw7PMtclNAmGT1eNDVfq6wJj5zsDMuLMZQ7VpppCX5gtMxQEcx\nG8uEF4qatShPWV4AZ4FCZEkST4XqhJxMLrS4wgfXp8fGa9uMgN98y4vDryb82bXleLRNpY/cJ+mh\nlw77e+96U+4NX3YulK80cTAqZ6E7XHAgtOg9xiVPN0eQiaNB+fxjZ6YQ8kwesblTDhjrJiu658Kr\nq0v6ZGOyKfXEfeTW9XTITkZ6n5dzJra8nXvJZPJQ6GLLYE/Y2z3auZqJYnTOfMWLkJWned9JfJ8K\nlpTysTBk94w/503cGTI9gWvFXa+9NhNowovenSeMwMrVH+yswp/7YS1yfxKRM+Z0QhDAvZ9klAJ3\nTZweXXDXCNyF7EgQcjuAtFes1ZeRYEBsLBgw3Gu6w7jmhybXU+zy/GN9QQyza9QO/3bksxIbgKNj\ngEKhEMaF+zfUhW5ttdna9nRvAqKFYp2Ijx9rc24nuFbec9UZVZemEleeA1j/fA2L3Fk168aeAOAV\nF53HV/7o4lhdfAwVk2oymboLfoT4ft3OZXQUclFOEcmErp/QNTB2GA7tiV4knt/azwQZroeYzUfW\npLdvyI43wtV/D5uD2Yi++2oiZpHHhVzqCHkkRF7dytUWOeXe8L5GFvmByCIuRvWho5/t63v585dF\nsc07jl/BmuLh6Lj+S3Gqt6hXtnDqv3sp+nXNFoKXyf6nAhHu6A/dSH5orguNfMW2aBmCmCswfA77\nwU6ndy/+o93rYvu6FAdhG/fcQm5+QK3e3pJyLpZ7372gfYv8GdMbnM/mBXe9Hrevj+9aiZWJwBjy\ny5SId0+O2rxBvkvVhR/qFH2Ag0/zTGYlTzMQuR8qhc7r2o1m7cNkcyvcmw0EMCMSDYjlyhygk3eN\nv5pLRt/NaHk55c6oMRzdGC0+8M9vPpcdxwYPqm+R57OZaF3GbAGK3RStDxjggMRdGQDHSeATPNwd\nWOTP20WSXb1y1/wvePOPogfVs7wGuusvLPDPbz43zN4WNqZcCWOvg58q1PnsV08EPRLpGeSY5b2R\nZVXhI98w0MnGNd62BicExYV8gLJ3fSj3RvXMFoL7dXhvIObOQrXifMgUGaUQpiUeHfdSk7pltAoJ\ns3EzWdh4aTjo6Re5nkWerXCt+IQhcrHehhVy32ru6ItcKyZPUY4Gs0h910q4bz//+Psv5LXnRLMN\nO8slSmP2JeW/IBrBlm3E5MN7HZ6vo69qP/Y9Ad0rIZOJwg89Ia+XQheILNll0QpMzqiYzJVj0SjH\nLfNi2CEu5FNEii0p5RnxLfLuwBD7jYna5jOmL37cfU8E7Suhx+Z6FaHQioRtNhrsbMwiH7VC7rtU\n92v4ocfhfTxvBzCllpB7jDqLfPvr4cpP8q18MMCRFYEuG1/cvRIQRijyx6+9kq/ddB45O7I/auJu\ng82rl/LyLcFAmfGUvLuUi1YL6hgAEbJCGP1xQDyL3D4ULi+H64ofrLDIyZdjyZti0TNSW1xcOTvs\nFOOwNyASvn0qfeQAfZPPBTndnVDYXgxd0QAqBC+tt7zEmwknjYUfxu5P5wAdhWwwOQtibpNgIdWO\nKKrCdYOt+Ow1S9i8egnfu3knV25Zy/uucoPd3jXxj1cD3301Id61rXjI67lWnF91Sou8tNRzrRQo\nc5TCuO9a8YXcE1eHf3zfem+EbDxUMCiPPa9/b92Le/iXoSj6PnLHIVMKX4IXnxxNqAnZ9jrY+jq4\n4qNA4I13cz1Mrhy+0JaW81Hu+ky1kDsLuVbTqvKR27rsITLCwjo7N+ehoeC5TzhoGErqBR24a+Ii\nzhp1rXT1Bm3WXwzG9Wpmw0c+EysEzS1H9nIoEzTCsDvvrDDPYh0qrGXZ0acY6rQDM7kinHoV2W/e\nZX9LJBBLVoPNFb/zpKjL9a6xV/PdydP4XMVNd+f1b8f6/k7GnnI++sAi8K3Ug56Qj9ju4ArZxyFT\npGQF99DoeP3FpX1BaSB7m2tYfvdObKn3mS5ufslJ/OAXzzHyuJ3lOrGH/XSSc/5OtzDy0uqIgZiw\nZHI186tEJxYqXSsHR8Z53lnklVZvrmSjOPBcK0E9nqObi05eQW9ngQ+8MkrkFfMVJ7lW6nDI6zFF\nURxBOF6mTnc6dK1kEiz6mB87G97bUfIMyP7gXoQWuSfOlVEnfpn8fa/9KjxZnT+9+rfBdffHacKX\nhe9iCC3yX8KGICWFs8j9l8BByuEz8PFXbWXcW0AbCH67IUppIRJNmDK5Uhj5sa7Pq6d7hn0ht+3w\nwEjyYha9HQXWLu8Hlx68O3ipXHbaKoYe6WGZDBMO6vrXtEZEk7uXR8e9+th773rUjbpWVixfxctO\nX8XTDz5StZsOdgIcfo5DueCNG4qHy9jmCcXtx7yXC0bfz+Fi/KY5cc1kPIu8hvV2+8RLedysqbI2\nQxeat1kENq6zD4X1KWdEwgfg+RoW+TBd4TTogyPj9X3Nvlg0IuS28foDmxP5oDz76eJNOwPfv2+R\nB/HOTsjt4gJLqxcujgt5tjHXim+hdi5DxFvW6+Bv4/smWeRW0EcpJIeB+T2WJoV8OJMwOHrlJ2H5\nyWTqHCsshy8UTiQ7qsM2IbjerjcW+siTrHAf/947Id9wHrzo5vq/g/C6xCxy5/7wB7LdtpHhUODd\nYix7M9F+h0wk5NmMJOZn9xGicNXJXClsX53+6laureWi6+gWTFnTk+zKKeQy/N0bo7Vb3XX56O9u\nYdktD/LDV9zNR64J1gON35/qwWSAlUtLYZ2iwgdl3Wet/LqulVy8Dazv7+BRExhBTy2NouBmw0ee\nLovcGDi8lyNdgSCE19vlUPasluHyOn5h4KKKB941omxGYP25sPpMOOcmLp4oelO541SKq7vRL9jQ\nz49+tY+n948gCKtWr4OnCR+ObEZ42gywhcfIZqK754RzQA7w0OQxYff1+dHxGpPtXUGatMjti8rv\n3h1esZ3up+4KHyxjosHO/snn+CmDFNw1cw9X5UxBqBhsy1IvKiuk2+vGl3t54/k9fKfvWvjq38Bg\nRdrbfAn22/wd7sHrDXzGXRypyvhYVaZ8c0J+INtbvfHky+Hky8kcrF7/1REaE75l61wrK21vwQ7i\nunZzxBSjlXmci6M38ocn0oprxRoq/mIV5BLGWLwoK1euciG4zk9P9uAClQ5Rqt9zrEAk8qv7Fnls\nScCEwc51fR187+ad9HVVL7wSlXMJXPt/Ig0Ity9l6+alhPKZydo4+dGq8FLHX/yHU9gy2MMLNngv\nVVuufXbeQq0IGiD+grCD2wfo4pLOf+Dlp66C3waGyWz4yNMl5Eefh8kxRm2WvLAtOcHxrBYnvpUi\n7BpRRoA1W+CGuwC4bX3t01ZqRnhMgXdcuon/9Pc/Cj47C78UPbifn3gRl2V/wL9nolW2/YGj/aaT\nLYOBiDw/OoVF7nev6wzAhdgG6/vDnz7/A3z/U+/gvslgIGrSmCiRE0fZbzpZ7opwxcfgzr9MHOGP\nWYiNuFYguj62/B0FuGzb8XDCQ9XiFOsKW2vQjhd0c7iGRW6vSa4Uv1YNsD/JIrc05P/3J+g46zqT\ngf/8aNg+3WGOUCAr9mF2rhV/LCSxEL5F3uRgpz1HzCI//Rp48HOw1kv65Pc8bBsuW6v51xO9uGZ7\niBKn18njnYQbCzHZcjj/IibkLkChwq002J8waF3JhurMpInkS4GQJ4WmAl3FHK8+e318oxXyvWIt\n8no9T18oyr1kbfbVsUwR4xkWpu195HbGmalsyKtt9+m0V4abnPVT+RBG3abGTYpKt4ET28lJEz++\nayAmmthz1+QZnDTyKZaVl4LNrT3ihUyduGEdS22X7uj4ZDwet5JmLXK7/6RXV9O1nHeOXxt+njQm\nlrdmmC5Wujqd/jvBvyR8C7HQWVPsrt4+yNd//AzXnLUOijUibZJ88K6bmslFrq/e4GV4rzmJLcck\nPIzuN026VQAOZ2oPjia5jc5Y18P9Tw4n/8AL8/Rfgu4aHY7FPvfE/8/WsD79e1+ceiA3hj12zEd+\n3E748/jakzEXoxNy6yMfcgOIXSv4zGvOYdPK5oTcxXubfJnDR4Nw0a5iwmzQpPGBmSJXAvbXtMgT\ncRa57dXWtch9yr1krJCLSDzUte0tchvznSsGDfKwix/uP66qUUYWefwQ7oI2M+BQaSWHngdjQktU\nhMgiNEG53Nu70qfrW+TjhaXkshmKuQyj45P1c5bEfOSNR4m4ldeT6jJpiAu5zdLX6LGDE3TV9JGv\nXFriGwm53KfEPdClnqiu5V7u3f017v9VkZcPJrhCnHU7DSHfekwfh4eKTGy+kkrHRVL6gc++YUfN\nQbhavSV3Xf3MnTHr8LpvJvd+IBL4XDlu/TeCPYd/nxNJsMjd+M04Obj+Dhg4gR3NCCF2sNNFrWQL\n4XyJmI/c1SnfgAU+XVzPvYZFnsipr4B/+SD7pAEfuU82H7abjMSNAfWRWyHPluyiwSPVK307QiGv\n8pEHn8cnpi/kTrwnTSTqIkCPdZ/0n1B1bn9qbyxzo3UTLSnnGTo4OoVrZYoZfJWs2wHAA5PRlOjK\n4x/T18HTT0Tl2U9nY/7PbDxCxC6LOTW739/YgxQKeXyx2+1nncP2sxL2B29maPNicMvujTy69TFO\nTnAZJPU2yoVs1QQx1p4FT91T8xyuPSRa5FA9TuDjRDZff/5A8omjPOx18d1bNuKm6PcQ121nOggS\nhu+JmeT50aCxdBQThDzJdz9TjFs/ejMvogv+BHbcyPBHHgRG6ketQDD4/NgdQPSsCRLLtqnhhzbP\nSsGuIHOwlkVEZMTV8pE3072pZdUbY+LHP/HiYCHc9ecF+3nfxS1yfypvIFSnrF7CnY8M1Xf41Op2\n1+LEi+EtP+XOv3qgZl3edcUp3HNcHr4SfA4s8gaO7Y/QF7rIjgQ/qutDBDjrDQ0cnJpCXhcn5NMQ\ng3w2kyjikOxaSeR1XwvzXdc7jhtcnsgUyTbqSnADxXWOXxNn1Ewl5AkWuTNart6eELnUxOldhlIx\n42xcGbwwzhz5LGznAAARUklEQVT087xYoTPTqF+juOyK5YTeXC0yGehaFk7+qxtHDnDBfwn+QRgA\nIAKXnrqKe8/ey6f/9ZcLL/xQRN4nIg+LyIMi8iURaa7P1SxHg4T5BbuCzIEjtS1yR+VDGFrklbGv\nTRzDfZyYNOH4Rri4w7HnhxaQ/7tY18q77OP5QDzOtK6CcLp5Es1GKwAsjWf5q3yxlQtZXrw5ekgD\nP2oDwuVbNZ5rxc8M2BLuRdGM9RQKeZOuhyloyNUEQS+ljsXsmsARm7ZhItvEC8cNFE9O3earyxW8\nOC45bQ2fvq5Wd4YKH3l03X/x33bzVy8/tfnzekzYyXIyOc7uU1fxvZt3xuZshIOd40cTfj1DuGu3\n4pT6+yX91Ipv3aRZFVSuY3u5zeC4EH3k3wJuMcaMi8h7gFuAP269WDWwFnmp3A0crmuROyqfwWYs\ncjcRsjIiw3WRJ73vkp71mGulhlXnkuu47Ip1i9VMzFcNEkXJs66DBRAaOFCla8Uet1Tpbpgu07HI\n3QScGRbyhi3yKXDX3lnkk1Mlu/LptpktpyPkW66FoYcZvOgdDNazRn2L3AsVbShFcR2EyEeeseWv\nika56C+CNUuPObulc9XlNV8OFpboPWbqfStwj+WUPU6P0LVi/z9pZTdfvPGFUVqCGaQlITfGfNP7\n+APgqtaKMwXWR75+1QDwKy7evLLmru5yV/o3T13bw52PDFWtEJPE1958Ht/9+VDV9jBqpdK1UkGu\nhkXuM1aomNw0yySexgvV6x9YwdreJn3MhU6ymaCHEyZBahVn2TYzMJUwqWQmcLfuwk01BiIbJFvh\nI3d56BvCzloMJ781Q7ELLv/I1PvFUhw0P2BcCxHhIROI5+jA5uSdVp4ShgLPGseeP+2fupDBZh7T\ncPKh/U1XMReGGs80M+kjvw74XK0vReQG4AaAwcHB6Z3BCvmK/j4e+cuTq1ZCTz5v/PNNu05g18bl\nnLZ2aoHYtGoJm1ZV+039qJV6ZGI+8uSyOiGfIaNvSqayrt7/mvOhXghkEoVOsnbRgNIUs/waxi0g\nXCuKI4ljzoHlm2HXnzR1KpcErRYiwvfffgF9nU2OUVTg2sNRm7dkPNvEC7PUAyfthm3XtVSG+UCA\nuybPZNfo+7jtxCvmuzjTwj3qDbvZqAiEmGWmFHIRuQNIMn1vNcZ82e5zKzAOfKbWcYwxtwG3AWzb\ntm16TiLrWiFfnnpacI2Ll80Ip69rzZUfxZHXD+6PZfasENDPjO9id/ZuRjvXxo45G9x764WJaQVi\ndC6HQ882J5yOTDZc9aQqkmO6uEGvjZc1/ptyD9z4/aZO88S7L21ov9U1pok3g7v2blbtk8ddTcOO\nIxG45rMtl2FKLvtQGHU1U7im/bhZ08TsjYXFNWcN8tE7H6Or2Ljt64ymWoujzyRTlsoYc2G970Xk\ntcBlwC4zG1OWfOxgZzPdvllZH89zrTiSXCN+zGll/Omt49dz6/j1fKE7sAYbmj0IcOJL4edfb6q8\nvhup5nlef0cwyagZn7THETvJozRTg527/gw2XQ6rTpt635Tg2sg3Jrdz547bOf/i/zjFL+aBWbb4\n58qFONO89eITefOuE+pP2KsgU2f8bKZpybUiIpcAbwNebIw5PDNFqsPYkUBsGgjDm80G47tW6r0n\n3I085/h+rjtnA//v0T2x799z5amhz6zh4l79mekNeFlqXpdpDABx4w+CTHlEq/TMmJB39MHxu2bm\nWAuMCbLsvOTy+S7GnOG3uXTKeFCHQq650jujaS7q3KqP/KNAEfiWvVk/MMb8fsulqsXYkWCyRxMi\nPRtdhKwXteJOkDiGaPe7aNMKdm6sdln8zvZorKBh10om21ielVo/n8lWtXxT8A840cYGtxJvrCx+\nUmqQT4uwrnNQ6VajVo6fqYI0xMRow7P2ZvPSOQtjYtKECYASww/dfg28TRpa83IGmC1f/JqecsP+\nZqV9mQt/8UKhMmplNklXPvJLPwBvfbipn8yGj/z45UEc6A0vqp+xLrTcbXD4vbdeyP1/elHivrM5\n2Dkf51GUJNqp+WVS5FqZexp1K8zi1VtazofW53ce/m3N/cLJR/Zt4gYd33n55qooiDmLI0/Xq1tR\nUksYtbLQXStpoP5w5Awcv46P3N3Iylmk175wffW+cxVH3k4mkbLgaKfmF8aRz8W55uAc88Jc+eJC\nIU9ooVdtXUs+K1x22qopj9Nw+GGLzNULQ1GSSGv44XRwz/RcGE+L3yKfXYO8Lsct6+LR/7q7oX3n\nqoGrRa7MJ+3U+sJnWgc7p89c6ZV7T7R6OnWtKO1AOzW/bGbuBjsXrZDPFdNJppPE3EWtzMlpFCWR\ndgo/nMtcKyrkM0ZrdyvtceSK0gjt1PwWVK6VtDJX7eVFJy7jpaes5B27N7V0nHlNY6soc0Q7Nb9w\nsHMOjLRFK+SO2c7jVcpn+firtrZ8nLmylNspakBZgLRR84vGOme/0ovWtZI2vZqr8ENFmU/ayUce\nhSbP/rkWrZA75jP8sBl0EFJpB9rJXplL6Vm0Qp62N7+6PJR2oJ1auXPrzoXbdNEKuSMlBrla5Epb\n0E4Gi7pWZgC35Fix2fUn54lWVypXlDTQTq08THE9B+datFEr15+7gdGxicQEVQsR1XGlHWineQz1\n8jDNNItWyEv5LH908UnzXYyGaacup9LGtFEzd0I+F0baohXytNFOlko7c+Gm5XQ2sRL7YqOdmrm3\nNPusn6t9W9QCQ+PI24NPXLt9voswr7RTK5+pPEyNkI6RwDZAdVxpB9rJhThZZ9GZmUaFfIGgrhWl\nHWinVr5tfS8Ab5hibd+ZQF0rC4S5yn6oKPNJO9krA13FcG3f2UblY4GgFrnSDqRtxnVamBEhF5G3\niogRkYGZOF47onHkSjug9srs0LKQi8g64GLgV60Xp31pp0EgRVFmlpmwyD8EvI30pDVZkGj4odIO\naDOfHVoSchG5Avi1MeaBBva9QUTuE5H7hoaGWjntokR95Eo7oD7y2WHKqBURuQNYmfDVrcA7CNwq\nU2KMuQ24DWDbtm1qvVegOq60A9rOZ4cphdwYc2HSdhE5FdgAPGD9u2uBfxORs4wxz8xoKdsAzX6o\ntAPaymeHaceRG2N+DCx3n0XkCWCbMWbPDJSr7ZhLHf/Y726hv6swdydUFIsO6s8OOiFogTCXPvJL\nT1s1Z+dSFB+V8dlhxoTcGLN+po7VjqihorQD2s5nB53ZuUDQ8EOlHVDXyuygQr5A0PBDRVGmiwr5\nAkGFXFGU6aJCvkAQvROKokwTlY8FglrkiqJMFxXyBYLOB1IUZbqokC8Q1CJXFGW6qJAvEFTIFUWZ\nLirkCwR1rSiKMl1UyBcIapErijJdVMgXCKrjiqJMFxXyBYJOXVYUZbqokCuKoqQcFXJFUZSUo0Ku\nKIqSclTIFUVRUo4KuaIoSspRIVcURUk5KuQLjJ6O/HwXQVGUlKGLLy8gvnTjC1nTU57vYiiKkjJU\nyBcQZw72zncRFEVJIepaURRFSTkq5IqiKCmnZSEXkT8QkYdF5Kci8t6ZKJSiKIrSOC35yEVkJ3AF\ncLoxZlREls9MsRRFUZRGadUifyPwbmPMKIAx5tnWi6QoiqI0Q6tCfiJwnojcLSLfFZHttXYUkRtE\n5D4RuW9oaKjF0yqKoiiOKV0rInIHsDLhq1vt7/uAHcB24B9E5FhjjKnc2RhzG3AbwLZt26q+VxRF\nUabHlEJujLmw1nci8kbgi1a47xGRSWAAUJNbURRljmjVtfJPwE4AETkRKAB7Wi2UoiiK0jitzuy8\nHbhdRH4CHAWuTXKrKIqiKLNHS0JujDkKvGqGyqIoiqJMA53ZqSiKknJUyBVFUVKOCrmiKErKUSFX\nFEVJOSrkiqIoKUeFXFEUJeWokCuKoqQcFXJFUZSUo0KuKIqSclTIFUVRUo4KuaIoSspRIVcURUk5\nKuSKoigpR4VcURQl5aiQK4qipBwVckVRlJSjQq4oipJyVMgVRVFSjgq5oihKylEhVxRFSTkq5Iqi\nKClHhVxRFCXlqJAriqKknFwrPxaRM4D/AZSAceBGY8w9M1EwRVEWD3/72u2MjE3MdzEWLS0JOfBe\n4J3GmK+LyG77+fyWS6UoyqJi58bl812ERU2rrhUDLLF/LwWebvF4iqIoSpO0apH/IfANEXk/wUvh\nhbV2FJEbgBsABgcHWzytoiiK4phSyEXkDmBlwle3AruAtxhjviAirwQ+CVyYdBxjzG3AbQDbtm0z\n0y6xoiiKEmNKITfGJAozgIj8HXCT/fiPwCdmqFyKoihKg7TqI38aeLH9+wLg0RaPpyiKojRJqz7y\nNwAfFpEcMIL1gSuKoihzR0tCboz5F2DrDJVFURRFmQY6s1NRFCXliDFzH0AiIkPAL6f58wFgzwwW\nZz5YDHWAxVEPrcPCYDHUAWa/HscYY5ZVbpwXIW8FEbnPGLNtvsvRCouhDrA46qF1WBgshjrA/NVD\nXSuKoigpR4VcURQl5aRRyG+b7wLMAIuhDrA46qF1WBgshjrAPNUjdT5yRVEUJU4aLXJFURTFQ4Vc\nURQl5aRKyEXkEhF5REQeE5G3z3d5aiEit4vIsyLyE29bn4h8S0Qetf/32u0iIn9j6/SgiGyZv5JH\niMg6EblTRB4SkZ+KyE12e2rqISIlEblHRB6wdXin3b5BRO62Zf2ciBTs9qL9/Jj9fv18lt9HRLIi\n8iMR+ar9nMY6PCEiPxaR+0XkPrstNe0JQER6ROTzIvKwiPxMRM5eCHVIjZCLSBb4GPBS4GTgGhE5\neX5LVZNPAZdUbHs78G1jzAnAt+1nCOpzgv13A/DxOSrjVIwDbzXGnAzsAN5kr3ea6jEKXGCMOR04\nA7hERHYA7wE+ZIw5HtgHXG/3vx7YZ7d/yO63ULgJ+Jn3OY11ANhpjDnDi7VOU3sC+DDwf40xG4HT\nCe7J/NfBGJOKf8DZwDe8z7cAt8x3ueqUdz3wE+/zI8Aq+/cq4BH79/8ErknabyH9A74MXJTWegAd\nwL8BLyCYeZerbFfAN4Cz7d85u58sgLKvJRCIC4CvApK2OtjyPAEMVGxLTXsiWAXt3yuv50KoQ2os\ncmAN8KT3+Sm7LS2sMMb8xv79DLDC/r3g62W752cCd5OyeliXxP3As8C3gMeBYWPMuN3FL2dYB/v9\nfqB/bkucyF8DbwMm7ed+0lcHCJaG/KaI/NCuGAbpak8bgCHgb62b6xMi0skCqEOahHzRYILXcyri\nPkWkC/gC8IfGmAP+d2mohzFmwhhzBoFVexawcZ6L1BQichnwrDHmh/NdlhngXGPMFgKXw5tE5EX+\nlyloTzlgC/BxY8yZwCEiNwowf3VIk5D/GljnfV5rt6WF34rIKgD7/7N2+4Ktl4jkCUT8M8aYL9rN\nqasHgDFmGLiTwA3RI0EOfYiXM6yD/X4p8NwcF7WSc4DLReQJ4H8TuFc+TLrqAIAx5tf2/2eBLxG8\nWNPUnp4CnjLG3G0/f55A2Oe9DmkS8nuBE+xofQG4GvjKPJepGb4CXGv/vpbA5+y2v8aOcO8A9nvd\ntHlDRIRgDdafGWM+6H2VmnqIyDIR6bF/lwl8/D8jEPSr7G6VdXB1uwr4jrWw5g1jzC3GmLXGmPUE\nbf47xpjfI0V1ABCRThHpdn8DFwM/IUXtyRjzDPCkiJxkN+0CHmIh1GE+Bw+mMdiwG/g5gZ/z1vku\nT51yfhb4DTBG8Ba/nsBP+W2C5fDuAPrsvkIQjfM48GNg23yX35brXIIu4oPA/fbf7jTVAzgN+JGt\nw0+AP7XbjwXuAR4jWGu2aLeX7OfH7PfHzncdKupzPvDVNNbBlvcB+++n7vlNU3uy5ToDuM+2qX8C\nehdCHXSKvqIoSspJk2tFURRFSUCFXFEUJeWokCuKoqQcFXJFUZSUo0KuKIqSclTIFUVRUo4KuaIo\nSsr5/+Nm8nd7li4TAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mLVRD2tmHxoJ",
        "colab_type": "text"
      },
      "source": [
        "# Part 3: Binary Classification"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Bndx3QO5HuR9",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Now turn the full data into numpy arrays\n",
        "\n",
        "# Not yet in the final \"X\" format!\n",
        "input_data = df[['open', 'high', 'low', 'close', 'volume']].values\n",
        "targets = df['Return'].values"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pwdN0u8mH_GN",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Now make the actual data which will go into the neural network\n",
        "T = 10 # the number of time steps to look at to make a prediction for the next day\n",
        "D = input_data.shape[1]\n",
        "N = len(input_data) - T # (e.g. if T=10 and you have 11 data points then you'd only have 1 sample)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8ivu7w3yIFcu",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# normalize the inputs\n",
        "Ntrain = len(input_data) * 2 // 3\n",
        "scaler = StandardScaler()\n",
        "scaler.fit(input_data[:Ntrain + T - 1])\n",
        "input_data = scaler.transform(input_data)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3S0TJMWHIG68",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Setup X_train and Y_train\n",
        "X_train = np.zeros((Ntrain, T, D))\n",
        "Y_train = np.zeros((Ntrain, 1))\n",
        "\n",
        "for t in range(Ntrain):\n",
        "  X_train[t, :, :] = input_data[t:t+T]\n",
        "  Y_train[t] = (targets[t+T] > 0)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "v5OPoVZnIIoq",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Setup X_test and Y_test\n",
        "X_test = np.zeros((N - Ntrain, T, D))\n",
        "Y_test = np.zeros((N - Ntrain, 1))\n",
        "\n",
        "for u in range(N - Ntrain):\n",
        "  # u counts from 0...(N - Ntrain)\n",
        "  # t counts from Ntrain...N\n",
        "  t = u + Ntrain\n",
        "  X_test[u, :, :] = input_data[t:t+T]\n",
        "  Y_test[u] = (targets[t+T] > 0)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TTXdPZW9ILDP",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "b5f4e8c1-7915-4e06-d5fd-abb3489b7b16"
      },
      "source": [
        "# make the RNN\n",
        "model = RNN(5, 50, 2, 1)\n",
        "model.to(device)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "RNN(\n",
              "  (rnn): LSTM(5, 50, num_layers=2, batch_first=True)\n",
              "  (fc): Linear(in_features=50, out_features=1, bias=True)\n",
              ")"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 44
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tkNDPzSgIPT8",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Loss and optimizer - it's classification now!\n",
        "criterion = nn.BCEWithLogitsLoss()\n",
        "optimizer = torch.optim.Adam(model.parameters(), lr=0.01)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kthbTpgqIab5",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Make inputs and targets\n",
        "X_train = torch.from_numpy(X_train.astype(np.float32))\n",
        "y_train = torch.from_numpy(Y_train.astype(np.float32))\n",
        "X_test = torch.from_numpy(X_test.astype(np.float32))\n",
        "y_test = torch.from_numpy(Y_test.astype(np.float32))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "acXZJ9gOIhTI",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# move data to GPU\n",
        "X_train, y_train = X_train.to(device), y_train.to(device)\n",
        "X_test, y_test = X_test.to(device), y_test.to(device)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "He5Z5-z_IjiZ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "7cb40cb5-87ed-4624-dc91-51bd5c613ae7"
      },
      "source": [
        "train_losses, test_losses = full_gd(model,\n",
        "                                    criterion,\n",
        "                                    optimizer,\n",
        "                                    X_train,\n",
        "                                    y_train,\n",
        "                                    X_test,\n",
        "                                    y_test,\n",
        "                                    epochs=300)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 5/300, Train Loss: 0.6917, Test Loss: 0.6930\n",
            "Epoch 10/300, Train Loss: 0.6915, Test Loss: 0.6936\n",
            "Epoch 15/300, Train Loss: 0.6898, Test Loss: 0.6913\n",
            "Epoch 20/300, Train Loss: 0.6860, Test Loss: 0.6925\n",
            "Epoch 25/300, Train Loss: 0.6830, Test Loss: 0.7031\n",
            "Epoch 30/300, Train Loss: 0.6794, Test Loss: 0.7024\n",
            "Epoch 35/300, Train Loss: 0.6727, Test Loss: 0.7206\n",
            "Epoch 40/300, Train Loss: 0.6663, Test Loss: 0.7369\n",
            "Epoch 45/300, Train Loss: 0.6622, Test Loss: 0.7610\n",
            "Epoch 50/300, Train Loss: 0.6486, Test Loss: 0.7806\n",
            "Epoch 55/300, Train Loss: 0.6330, Test Loss: 0.7904\n",
            "Epoch 60/300, Train Loss: 0.6125, Test Loss: 0.8240\n",
            "Epoch 65/300, Train Loss: 0.5842, Test Loss: 0.9422\n",
            "Epoch 70/300, Train Loss: 0.5465, Test Loss: 1.0212\n",
            "Epoch 75/300, Train Loss: 0.5248, Test Loss: 1.0939\n",
            "Epoch 80/300, Train Loss: 0.4848, Test Loss: 1.1769\n",
            "Epoch 85/300, Train Loss: 0.4510, Test Loss: 1.2732\n",
            "Epoch 90/300, Train Loss: 0.4160, Test Loss: 1.3699\n",
            "Epoch 95/300, Train Loss: 0.3899, Test Loss: 1.4148\n",
            "Epoch 100/300, Train Loss: 0.3897, Test Loss: 1.4258\n",
            "Epoch 105/300, Train Loss: 0.3455, Test Loss: 1.5853\n",
            "Epoch 110/300, Train Loss: 0.3030, Test Loss: 1.7515\n",
            "Epoch 115/300, Train Loss: 0.2688, Test Loss: 1.8953\n",
            "Epoch 120/300, Train Loss: 0.2549, Test Loss: 2.0888\n",
            "Epoch 125/300, Train Loss: 0.2352, Test Loss: 2.1472\n",
            "Epoch 130/300, Train Loss: 0.2183, Test Loss: 2.2414\n",
            "Epoch 135/300, Train Loss: 0.2019, Test Loss: 2.4252\n",
            "Epoch 140/300, Train Loss: 0.1625, Test Loss: 2.5628\n",
            "Epoch 145/300, Train Loss: 0.2348, Test Loss: 2.2261\n",
            "Epoch 150/300, Train Loss: 0.3668, Test Loss: 2.0730\n",
            "Epoch 155/300, Train Loss: 0.2386, Test Loss: 2.0185\n",
            "Epoch 160/300, Train Loss: 0.1679, Test Loss: 2.1199\n",
            "Epoch 165/300, Train Loss: 0.1215, Test Loss: 2.2486\n",
            "Epoch 170/300, Train Loss: 0.0901, Test Loss: 2.4675\n",
            "Epoch 175/300, Train Loss: 0.0647, Test Loss: 2.6855\n",
            "Epoch 180/300, Train Loss: 0.0457, Test Loss: 2.8607\n",
            "Epoch 185/300, Train Loss: 0.0322, Test Loss: 3.0115\n",
            "Epoch 190/300, Train Loss: 0.0217, Test Loss: 3.1653\n",
            "Epoch 195/300, Train Loss: 0.0155, Test Loss: 3.3461\n",
            "Epoch 200/300, Train Loss: 0.0115, Test Loss: 3.4587\n",
            "Epoch 205/300, Train Loss: 0.0085, Test Loss: 3.5405\n",
            "Epoch 210/300, Train Loss: 0.0066, Test Loss: 3.6208\n",
            "Epoch 215/300, Train Loss: 0.0053, Test Loss: 3.6755\n",
            "Epoch 220/300, Train Loss: 0.0044, Test Loss: 3.7326\n",
            "Epoch 225/300, Train Loss: 0.0037, Test Loss: 3.7751\n",
            "Epoch 230/300, Train Loss: 0.0032, Test Loss: 3.8169\n",
            "Epoch 235/300, Train Loss: 0.0028, Test Loss: 3.8485\n",
            "Epoch 240/300, Train Loss: 0.0025, Test Loss: 3.8741\n",
            "Epoch 245/300, Train Loss: 0.0022, Test Loss: 3.8990\n",
            "Epoch 250/300, Train Loss: 0.0020, Test Loss: 3.9220\n",
            "Epoch 255/300, Train Loss: 0.0019, Test Loss: 3.9445\n",
            "Epoch 260/300, Train Loss: 0.0017, Test Loss: 3.9654\n",
            "Epoch 265/300, Train Loss: 0.0016, Test Loss: 3.9853\n",
            "Epoch 270/300, Train Loss: 0.0015, Test Loss: 4.0042\n",
            "Epoch 275/300, Train Loss: 0.0014, Test Loss: 4.0221\n",
            "Epoch 280/300, Train Loss: 0.0013, Test Loss: 4.0388\n",
            "Epoch 285/300, Train Loss: 0.0012, Test Loss: 4.0553\n",
            "Epoch 290/300, Train Loss: 0.0011, Test Loss: 4.0713\n",
            "Epoch 295/300, Train Loss: 0.0011, Test Loss: 4.0866\n",
            "Epoch 300/300, Train Loss: 0.0010, Test Loss: 4.1014\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NzD5qsMxJxuY",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "02733c51-d842-4404-b746-ad0087b12154"
      },
      "source": [
        "# Plot the train loss and test loss per iteration\n",
        "plt.plot(train_losses, label='train loss')\n",
        "plt.plot(test_losses, label='test loss')\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWoAAAD4CAYAAADFAawfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngloIITycBLUQQni4/w9LNUIN/FhNtgAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5IACnu3gJ4dK",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "b74b4960-2b57-4d9c-bdad-e19f86236cf1"
      },
      "source": [
        "# Get accuracy\n",
        "with torch.no_grad():\n",
        "  p_train = model(X_train)\n",
        "  p_train = (p_train.cpu().numpy() > 0)\n",
        "  train_acc = np.mean(y_train.cpu().numpy() == p_train)\n",
        "\n",
        "  p_test = model(X_test)\n",
        "  p_test = (p_test.cpu().numpy() > 0)\n",
        "  test_acc = np.mean(y_test.cpu().numpy() == p_test)\n",
        "print(f\"Train acc: {train_acc:.4f}, Test acc: {test_acc:.4f}\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train acc: 1.0000, Test acc: 0.4902\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}