JTAFinance (Team 3)

Inspiration

The inspiration behind this project is the cost-of-living and household debt crisis, especially in experience cities like Toronto.

What it does

Calculates baseline financial risks for each age group, then simulates financial shock scenarios like increased inflation and decreased income to identify which age group(s) would have the biggest increase in financial vulnerability post-shock. A RandomForest classifier was used and a permutation test was ran on the dataset to identify which feature(s) are the most useful in predicting financial risk, to see how age group compares with other features in prediction.

How we built it

We analysed the data in VSCode using Python, utilising Python libraries such as NumPy, pandas, and MatPlotLib to do calculations and visualise the graphs.

Challenges we ran into

A challenge we ran into was in relation to variable bias when using RandomForest. At first, we used a RandomForest model to calculate importance metrics to determine which features are most tied to financial risk, but found out that RandomForest importance metrics often favor numerical data like "Income" over categorical data like "Province." To ensure a fair analysis, we pivoted to running a permutation feature analysis, which allowed us to "stress-test" the model by shuffling data to find the true, unbiased drivers of risk.

Accomplishments that we're proud of

We are proud that we ran 4 different shock scenarios to identify increased vulnerability amongst age groups, and were able to tie that in with financial risk amongst household groups because of the typical household responsibilities of age groups (e.g. younger people tend to rent, older people tend to own houses). We are also proud of being able to have a wider perspective on at risk rate vs shock severity with the stress test, extending it to all of Canada and seeing the difference amongst age groups.

What we learned

From the data, we learned that total liquidity is the most important feature when it comes to predicting risk. From doing this project, we learned how to utilise different Python libraries to analyse and visualise data.

What's next for JTAFinance

We would want to implement real-time integration, where we would feed live CPI (Inflation) and Interest Rate data into our Fragility Curves to provide a real-time financial report.

Built With

• jupyter
• python
• vscode