In this project, we evaluate different machine learning approaches to predict mortgage application decision. This is classification task with 8 classes. Our data set is public data related to Washington State home loans. Each year,these data are updated by banks and other financial institutions. The prediction methods used in this project are: Naive Bayes, Random Forestand Support Vector Machines. We compare these machine learning methods and results, and select the best one by accuracy of prediction.
The steps are described in instructions.txt file and R script, and results are interpreted in the report.
Naive-Bayes: Accuracy of 63%
Random Forest with 400 trees: Accuracy of 66%
Support Vector Machines with radial kernel, cost 10 and gamma 0.05: Accuracy of 67%
Official dataset can be found on: https://www.kaggle.com/miker400/washington-state-home-mortgage-hdma2016