Before the prediction, understand the data.
Financial ratios tell one part of a company's story. Macroeconomic conditions tell another. This research brings them together to investigate financial distress among Indonesian listed companies.
Explore the projectDefine the early-warning question.
Can financial ratios and macroeconomic conditions help identify financial distress? Start with the question, not the classifier.
Return on assets describes profitability relative to the asset base. Example series, not reported research results.
The chart uses illustrative values to explain data relationships. It does not reproduce reported results, run a classifier or recommend an investment.
How does a research question become a defensible model?
Follow the PPDAC cycle from the question to validation. The repository contains financial and macro datasets, exploratory analysis and comparisons using Logistic Regression, Random Forest and XGBoost.
A research pipeline that follows the PPDAC cycle — Problem, Plan, Data, Analysis, Conclusion — and uses XGBoost to detect early signs of financial distress.
Problem before algorithm
PPDAC keeps the work organized around the research question: can financial and macroeconomic signals support an early warning for distress?
Respect time
The implemented research notebook uses a temporal split: earlier observations for training and later observations for evaluation. Randomly mixing years would answer a different question.
Compare and inspect
The notebook compares a linear baseline and tree-based models, then examines evaluation and feature importance. The portfolio does not turn that research into investment advice or a live prediction.
My contribution
- PPDAC research framing
- XGBoost classification pipeline
Built with
- Python
- XGBoost