gbwalker/water_point

A stacked ensemble model for classifying potable water sources in Tanzania. # A Stacked Ensemble Model to Classify Potable Water Sources in Tanzania This algorithm employs a stacked ensemble machine learning model to classify the functionality of OOF potable water sources across Tanzania. Implemented for API 222: Machine Learning and Big Data Analytics (Harvard Kennedy School, April 2019). * `Course Competition.pdf` is a description of the assignment. * `walker_memo` provides an introduction and description of the submitted model. * `improved.R` contains the final model with a much stronger KNN function within the ensemble. Note that this takes much longer to run than the submitted version. * `submission.R` contains the submitted model with a weak KNN function. * `test.csv` is the OOF set and `training.csv` is the training set. * `walker_prediction_new.csv` contains the improved model's output. * `walker_prediction.csv` contains the submitted model's output.