HeyThatsViv/Functional-Status-of-Water-Wells

Project using machine learning to predict if water wells in Tanzania are functional, non functional, or in need of repair. Written with python using jupyter notebook for the main project flow/analysis and some visual studio code. # Predicting Functional Status Of Water Wells In Tanzania **Author**: Vivienne DiFrancesco The contents of this repository detail an analysis of classification of Tanzanian water wells as either functional, non functional, or needs repair. This analysis is detailed in hopes of making the work accessible and replicable. A Tableau Public story was created in companion with this project and can be found here: public.tableau.com ## Repository Structure - README.md: The top level README for reviewers of this project - main_notebook.ipynb: narritive documentation of analysis in jupyter notebook - TanzaniaWaterWellsSlides.pdf: pdf version of project presentation slides - Data folder: Contains datasets used in this project ## Business problem The purpose of this project is to use machine learning classification models to predict the functional status of water wells in Tanzania. The different status groups for classification are functional, non functional, and functional but needs repair. The hope is that by predicting the functional status of a well, access to water could be improved across Tanzania. ## Data The data used for this project is from the Data Driven website where, at the time of completing this project, it is an active competition. The dataset contains nearly 60,000 records of water wells across Tanzania. Each record has information that includes various location data, technical specifications of the well, information about the water, etc. The website provides a list of the features contained and a brief description of each. The link to the website to obtain the data for yourself is: drivendata.org You can also get the .csv files from the "Data" folder of this repository. ## Approach The approach for this project was to create many different model types to see what performs the best and to compare and contrast the different types of models. The way the data w …