georgerichardson/pumpit

Predicting water pump failure in Tanzania for Driven Data Pump it Up challenge # DrivenData Pump It Up Challenge Predicting water pump failure for Driven Data Pump it Up challenge. **This is a work in progress** **Current best classification rate: 0.8131** ## Introduction The Pump It Up challenge provides training data and the current operating status for over 50,000 water pumps in Tanzania. The objective is to use machine learning classifiers to predict the unknown status of a pump, given features that are known about it. This repo contains Jupyter notebooks to load, clean, and explore training data for the challenge and then to train machine learning algorithms to make predictions. In addition to the notebooks, there are also helper functions in `scripts/pumpitup.py` that provide useful functionality. Have fun exploring the notebooks or cloning the repo and playing with them yourself.