Merge pull request #50 from transferwise/tree_depth

Correctly handle max_depth in tree solver

wise_pizza/slicer.py

@@ -195,6 +195,7 @@ def fit(
                 dims=dims,
                 time_basis=self.time_basis,
                 num_leaves=max_segments,
+                max_depth=max_depth,
             )
             self.nonzeros = np.array(range(self.X.shape[1]))
             Xw = csc_matrix(diags(self.weights) @ self.X)

wise_pizza/solve/tree.py

@@ -14,7 +14,7 @@ def tree_solver(
     dim_df: pd.DataFrame,
     dims: List[str],
     time_basis: Optional[pd.DataFrame] = None,
-    max_depth: int = 3,
+    max_depth: Optional[int] = None,
     num_leaves: Optional[int] = None,
 ):
     if time_basis is None:
@@ -27,7 +27,7 @@ def tree_solver(
     df["__avg"] = df["totals"] / df["weights"]
     df["__avg"] = df["__avg"].fillna(df["__avg"].mean())
 
-    root = ModelNode(df=df, fitter=fitter, dims=dims)
+    root = ModelNode(df=df, fitter=fitter, dims=dims, max_depth=max_depth)
 
     build_tree(root=root, num_leaves=num_leaves, max_depth=max_depth)
 
@@ -66,19 +66,23 @@ def __init__(
         df: pd.DataFrame,
         fitter: Fitter,
         dims: List[str],
+        max_depth: Optional[int] = None,
         dim_split: Optional[Dict[str, List]] = None,
-        depth: int = 0,
     ):
         self.df = df.copy()
         self.fitter = fitter
         self.dims = dims
+        self.max_depth = max_depth
         self._best_submodels = None
         self._error_improvement = float("-inf")
         self.children = None
         self.dim_split = dim_split or {}
-        self.depth = depth
         self.model = None
 
+    @property
+    def depth(self):
+        return len(self.dim_split)
+
     @property
     def error(self):
         if self.model is None:
@@ -94,9 +98,19 @@ def error(self):
 
     @property
     def error_improvement(self):
+        if self.max_depth is None:
+            self.max_depth = float("inf")
         if self._best_submodels is None:
             best_error = float("inf")
-            for dim in self.dims:
+
+            if self.depth > self.max_depth:
+                raise ValueError("Max depth exceeded")
+            elif self.depth == self.max_depth:
+                iter_dims = list(self.dim_split.keys())
+            else:
+                iter_dims = self.dims
+
+            for dim in iter_dims:
                 if len(self.df[dim].unique()) == 1:
                     continue
                 enc_map = target_encode(self.df, dim)
@@ -122,14 +136,14 @@ def error_improvement(self):
                         fitter=self.fitter,
                         dims=self.dims,
                         dim_split={**self.dim_split, **{dim: dim_values1}},
-                        depth=self.depth + 1,
+                        max_depth=self.max_depth,
                     )
                     right_candidate = ModelNode(
                         df=right,
                         fitter=self.fitter,
                         dims=self.dims,
                         dim_split={**self.dim_split, **{dim: dim_values2}},
-                        depth=self.depth + 1,
+                        max_depth=self.max_depth,
                     )
 
                     err = left_candidate.error + right_candidate.error
@@ -141,13 +155,6 @@ def error_improvement(self):
         return self._error_improvement
 
 
-def mod_improvement(improvement: float, depth: int, max_depth: int) -> float:
-    if depth < max_depth:
-        return improvement
-    else:
-        return float("-inf")
-
-
 def get_best_subtree_result(
     node: ModelNode, max_depth: Optional[int] = 1000
 ) -> ModelNode:
@@ -156,8 +163,8 @@ def get_best_subtree_result(
     else:
         node1 = get_best_subtree_result(node.children[0])
         node2 = get_best_subtree_result(node.children[1])
-        improvement1 = mod_improvement(node1.error_improvement, node1.depth, max_depth)
-        improvement2 = mod_improvement(node2.error_improvement, node2.depth, max_depth)
+        improvement1 = node1.error_improvement
+        improvement2 = node2.error_improvement
         if improvement1 > improvement2:
             return node1
         else: