An ML-approach to predicting bank failure. A part of the requirements for CPTR 435 Machine Learning and ECON 328 Money and Banking.
Data is to be retrieved from the FDIC using their BankFind API from https://banks.data.fdic.gov/docs/#/Structure/searchInstitutions
This project is based on work from:
https://doi.org/10.1016/j.ribaf.2017.07.104 - Predicting bank failure: An improvement by implementing a machine-learning approach to classical financial ratios
https://doi.org/10.1016/j.dss.2012.11.015 - Partial Least Square Discriminant Analysis for bankruptcy prediction
https://doi.org/10.1016/j.eswa.2008.01.053 - Effects of feature construction on classification performance: An empirical study in bank failure prediction
This project involves the use of binary classification to predict bank survival/failure based on its financials. Features consist of accounting values and ratios reported to the FDIC, and the target is a binary variable indicating the bank's survival (1 = active, 0 = failed). We aim to reproduce the features described by Le, H.H., & Viviani, J.H. (2018), which is the first article linked above.
Step 1: Develop the API GET calls to pull the required financials from the FDIC database. The features to be used in the model are described in the literature linked above. We can retrieve them by following the guide to variables provided by the FDIC in the risview_properties.yaml file.
Step 2: Clean and preprocess the data as necessary.
Step 3: Train, test, and evaluate several competing ML models for binary classification.