PaulVNarducci/Proj_3_Tanzanian_Water_Wells

Project for the Flatiron school in which I determined the functionality of water wells in Tanzania. Data comes from Driven Data competition which I also took part in. # Module 3 Project - Tanzanian Water Wells ## Introduction Tanzania is a country located in East Africa in a region known as the Great Lakes Region. Currently, Tanzania is having a water crisis in which 30 million people, out of a population of roughly 57 million people, lack access to a sanitary source of water. In this project, I will be using various machine learning models on data from Driven Data's Pump It Up: Data Mining the Water Table Competitionin order to determine the functionality of water wells in Tanzania. The purpose of this competition is to determine what wells are functional, functional, but need repairs, and non-functional in order to improve maintenance operations so that clean and useable water is available to the people of Tanzania. Since the dataset from Driven Data is a multiclassification data set, I decided to work with three different machine learning models in order to determine which was best at producing the strongest model for this competition. I used k-Nearest Neighbor, Random Forest, and XGBoost in order to determine which machine model would produce the most viable performance. After running my models, I tested their accuracy by uploading my model results to the Driven Data competition. The following results were produced: For my purposes, I will only be discussing my data in terms of the Random Forest Classifier model that I created because it produced a score of .8081 when compared to my k-Nearest Neighbors and XGBoost Classifier models. ## Results In order to determine what features had the most significant impact on my RF model, I used the .feature_importances_ method in order to create a bar graph of the top 20 important features. According to the graph, Latitude and Longitude were the two major factors used when determining the best value for this model. The map may show where these wells may be located, but it does not explain why the wells are located in these areas. When looking at the next most important feat …