sean2108 / bankruptcyprediction

BC3409 Project: Bankruptcy Prediction

Running the code

Remember to add the datasets into the ./data/ folder. Or change the directory written in the code. Please run simfinmodel.ipynb to run our LSTM/GRU models to predict bankruptcy on our dataset.

Problem Statement

Using AI and deep learning to perform financial evaluation of firms (with a focus on Bankruptcy Prediction) via big data.

Dataset

Our main dataset is obtained from SimFin

• 2165 companies
  • List of company names can be found in `listofcompanies.txt'
    • Mylan N.V. is duplicated
  • Out of these, 49 companies are bankrupt / had prior history of bankruptcy, either in its subsidaries or the company as a whole
• 10 years of data ranging from 1/1/09 to 11/1/19
• 42 attributes of financial data (including financial information of company and market ratios)

Our secondary dataset is obtained from Machine Learning Repository

• 5190 companies
  • 1 year of data ranging from 1/1/12 to 12/1/12
  • 64 attributes of financial data (different from our main dataset)

Our final dataset used after preprocessing based on our main dataset is data_with_ratios.pickle.