Team 47 Personal Finance Case- Long Term Stability

• 

  SHAP Graph

Inspiration

Toronto is one of the most expensive cities in Canada, with individuals facing rising rent, tuition, and basic living expenses. As a result, many Canadians are concerned about their financial well-being.

As of 2025, approximately 49% of workers cite financial stability as their primary stressor, with 40% constantly worrying about their future (Benefits Canada). This stress directly reduces workplace productivity & has a significant impact on mental health.

We are trying to understand the primary causes of this financial stress and what's impacting long-term stability, to raise awareness and help address these factors.

What it does

Quantifying which financial and demographic variables most strongly predict long-term stability.

Our system builds a hybrid machine learning pipeline that stacks Elastic Net Logistic Regression and XGBoost to predict factors impacting long term stability, handles missing values with median imputation, and scales features for the logistic model. It fits the models on training data, generates predictions on the test set, and is structured so one can later analyze factor importance or interpret coefficients.

How we built it

We took the primary indicators of financial stress in the long term and ensured the data was structured and cleaned. Then, we used a stack of Logistic Models with Regularization + Gradient Boosting (XGBoost).

Other technologies used: Python, Pandas, NumPy, Google Colab, xgboost, shap, scikit-learn

Challenges we ran into

The data was skewed, so we ran into issues trying to reduce the skewness. Additionally, we faced a challenge in stacking the two models, and specifically, not being able to utilize a complex neural network for the task.

Accomplishments that we're proud of

We are proud to have reduced data skewness and mastered logistic models with regularization alongside Gradient Boosting (concepts we had limited knowledge about before).

What we learned

We learned a lot about how to master stacking two different ML models with each other, and also understanding the theory behind logistic models with regularization & Gradient Boosting.

What's next for Personal Finance Case- Long Term Stability

Work on reducing data skewness and having a more complex internal neural network with more indicators and weights to analyze the primary causes of long-term stability.

Built With

• excel
• google-colab
• numpy
• pandas
• python
• scikit-learn
• shap
• xgboost