tvrjuana/tanzania-water-wells

# World Wide Water - Predicting Well Functionality in Tanzania for UNICEF World Wide Water analyzed thousands of water wells in Tanzania to find what attributes influence well failure the most. With our analysis, stakeholders like UNICEF, can gain better understanding about Tanzania's ongoing water crisis. ## Organizational Objective As it stands, much of the Tanzanian population lacks a basic access to consumable water. Many Tanzanian citizens collect water from groundwater sources or turn to private lenders to meet their needs. However, both options are risky and endanger consumers to illnesses like cholera and malaria. It is imperative for those who seek to improve water access in Tanzania to understand what factors are most influential in water well failure. ## Main Findings We analyzed over fifty-four thousand water wells built between 1960-2013. We found three features that powerfully impact the functionality of water wells: 1. Age of Well 2. Location of Well 3. Water Extraction Method Our final model garnered a recall score of 78%, displaying a powerful ability to correctly categorize functional vs. non-functional wells. ## The Notebooks The technical lead operated out of the mackoy branch, while technical associates assisted through pair programming. The main files for this project are in main_notebook.ipynb. ## Data Cleaning The dataset, tanzania_wells.csv, was is the primary data source. The dataset contained comprehensive information including but not limited to water quantity, latitude, longitude, donor information, population metrics, and extraction type. We featured engineered information for the age of the well. After a robust data cleaning effort and initial statistical analysis, the team modeled the data. ## Model Analysis ### EDA ### Well Age The first analysis concerns the relationship between the percentage of non-functional water wells and the age of the respective water well. We found that older wells perform worse than wells that a …