HACKATON PROJECT: Predicting provided delivery time frames by TNT

This is a work in progress, at this moment the prediction accuracy is 80%.

How to run a server locally

1. Open the project with RStudio.
2. Go to your console and run the following commands:

> library(plumber)
> p <- plumb("plumber.R")
> p$run(port = 5762)

3. Then you can access to this portal locally http://127.0.0.1:5762/__swagger__/

TODOS

A list of the first things that come to my mind at the moment, probably there are gonna be more:

1. Refactor the plumber.R file, at this moment that file is handling everything. The data normalization and the KNN should be on a different place.
2. The list of suburbs that we have at the moment are the suburbs that have been used by TNT shipments. Any suburb that has not been used by TNT it won't appear on this list. However, this list is limited from 1/January/2018 and also we removed shipments coming from big companies.
3. Create a well balanced dataset with a normal distribution of shipments. This means to get a uniform list of all the states using the majority of the suburbs and postcodes.
4. Improve the endpoint, we just have the echo endpoint.
5. Add authentication to the web service.
6. Handle request errors properly.
7. Show better errors as a web service.
8. And more to go 😊.