andrewschmeck/Modeling-Water-Wells-in-Tanzania

Classification Project for NGO's # Modeling Water Wells in Tanzania Flatiron Phase 3 Project, by Kelsey Lane, Andy Schmeck and Ted Brandon This repository has been created to present a data analytical approach to help an NGO predict failure of water wells in Tanzania. This document is intended to bridge the gap between technical and non-technical audiences. ## Repository Contents: Technical Jupyter Notebook (.ipynb) containing all data analysis Matching pdf of the above notebook Phase 3 presentation (pdf) Phase 3 presentation slides ## Overview: This project uses data about Tanzanian wells available at DrivenData. Our model seeks to classify wells as functional or not, in order to help a non-government organization (NGO) determine which wells they need to repair. After the data were cleaned, various iterative classification models were run in order to identify a model that has the best mix of a high precision and low complexity. We found that the amount of total static head (tsh) of the well, alongside its installer, extraction type, and if it had a government permit, were all significant predictors in classifying the functionality of a well, with 75% precision. ## Business Problem An NGO seeks to help Tanzania accomplish its Millenium Development Goal 7C: "halving the proportion of the population without sustainable access to safe drinking water," (MDG report). Tanzania's situation is dire (World Bank): - Only 60\% of Tanzanians get their drinking water from an improved source. - Of the 83,000 rural water points recorded in the national water point census, as of 2014, 40\% were found to be non-functional, with the likelihood of failure highest at 20\%, their first year of operation. Improving water supply, sanitation, and hygeine conditions have been linked with improved human development, reduced poverty, and reduced stunting in early childhood (World Bank). 46% of the wells in Tanzania are in need of repair or nonfunctioning. Instead of building new wells, an NGO can drastically increa …