This project was done for the University of Pennsylvania Data Analytics Bootcamp using Python and the libraries Pandas and Scikit-Learn working in a Jupyter Notebook. The goal of the project was to explore how two different supervised machine learning models would perform when using loan data to predict credit risk. The first step in this project was to prepare the data by dropping an unneeded column from the training data to match the test data, converting categorical to numeric using Get Dummies for both the training and test data, and separating the target from the features for both the training and test data. After preparing the data, two models were trialed: Logistic Regression and Random Forest. After the first round of training, testing, and scoring, the data was scaled with Standard Scaler and the models were trained, tested, and scored again on scaled data. The Random Forest model initially performed better but after scaling the data the Logigistic Regression model turned out to be better. This seems to indicate that when working with this type of data and a Logistic Regression model, scaling the data is key.
twolightsabovethesea / predicting-credit-risk Goto Github PK
View Code? Open in Web Editor NEWComparing the use of two different machine learning models to predict credit risk based on loan data