This repository contains a predictive analytics term project for a graduate course, Predictive Analytics, at Bellevue University. The group members of this project were Sam Sears, Jolene Branch, and Andrea Fox. The goal of the project was to predict hospital readmissions for diabetic patients using the Diabetes 130-US hospitals for years 1999-2008 Data Set from the UCI Machine Learning Repository (https://archive.ics.uci.edu/ml/datasets/diabetes+130-us+hospitals+for+years+1999-2008).
There are 4 main notebooks in the repository that contain data preparation, modeling, and modeling results; initial modeling, updated modeling, random forest using H2O, and cost sensitive random forest using costcla. The Updated Modeling notebook contains the best performing model in terms of business costs, sklearn's random forest. Data exploration notebooks containing analysis of correlations and readmission rates by variable can be found in the Data Exploration folder.