Case study
Stock Predictor
A machine learning price-movement classifier. Decision Tree model in Python with engineered features, preprocessing pipeline, and Jupyter-based evaluation. Coursework project, not production.
- Role
- Personal project
- Period
- 2025-04 → 2025-05
- Status
- Shipped
- Tags
- 4
Summary
A Decision Tree classifier over engineered market features. Built the preprocessing and evaluation flow in Jupyter notebooks. Coursework, not production. Did not make money. That was never the goal.
What I built
- Engineered market features from raw price data.
- A preprocessing pipeline and a Decision Tree classifier in Python with scikit-learn.
- An evaluation flow in Jupyter notebooks to inspect per-class results and feature importance.
What was hard
Not the model. The data work: cleaning the inputs, deciding what a "feature" actually meant in this domain, and writing the evaluation so the numbers were honest about what the classifier could and could not do.
What's next
The notebooks and code are in the repo. This was coursework and the classifier is not a trading system.
What I learned
The goal was to see how much of a real ML pipeline I could wire up end-to-end without hand-waving the data work.