Regression and Classification Metrics Dashboard for Model Evaluation

Model evaluation is where data science becomes accountable: a single accuracy number hides more than it reveals, and practitioners need fluency across regression and classification metrics to judge a model honestly. This project builds the tool that makes those metrics concrete and comparable.

The intern creates a Streamlit dashboard where users upload model predictions alongside true values and instantly receive the appropriate evaluation suite. For regression tasks the app computes MAE, MSE, and R-squared with pandas and scikit-learn, pairing the numbers with plots that show how errors are distributed; for classification it renders an annotated confusion matrix and derived metrics so users can see exactly where a model confuses one class for another. The dashboard validates uploads gracefully before computing anything, handling mismatched lengths and malformed files the way a real tool must.

The project drills the vocabulary every data scientist is tested on — what each metric measures, when it misleads, and how to read a confusion matrix at a glance — while giving the intern experience building a small, genuinely reusable analytics utility they can point at the outputs of any future model.