Also available on GitHub Pages: https://ucb-stat-159-s22.github.io/hw07-Group11/
Note: This repository is public. The Credit Card Fraud Detection Dataset is from Kaggle. The EDA and Logistic Regression Model were developed by Joe, Isaac, and Uma as a homework assignment for the Spring 2022 installment of UC Berkeley's Stat 159/259 course, _Reproducible and Collaborative Data Science.
In our studies, we will take a look at credit risk and how default status varies within different groups of people with different characteristics. We will not only try to find the relationship of some of our selected variables like how does default rate varies among gender or income classes, but will also build and train a simple logistic regression model that can help us predict whether a client is likely to default or not. We are curious about this topic because we believe this is actually a very import question to solve in the real-life financial world, and we wonder what how good can we predict if one is actually going to default or not. The entire analysis is contained in main.ipynb with computation details contained in details.ipynb
The following are the available make commands:
env: Creates the environment and configures it by activating it and installing ipykernel into ithw7_tools: Installs the hw07 tools packageall: Executes and generates outputsclean: Removes all figures from the /output directory
Install the packages:
pip install . or make hw7_tools
Test the packages:
pytest hw7_tools
After make clean, rerun details.ipynb or make all to regenerate the data and outputs needed for main.ipynb
Detailed data computations of processing/plotting/modeling/serialization can be found in details.ipynb, if you are curious and want to play with our results(e.g. by adjusting data/parameters), please also look into details.ipynb and do not work in main.ipynb since all outputs are loaded into main.ipynb instead of being computed there.
The project is released under the BSD 3-clause License.