Overview
Food Item Classifier predicted food items from student responses and compared multiple model families to understand generalization performance.
Motivation
The goal was not just to train a model. The useful part was comparing how different approaches behaved under the same prediction task and what kinds of features improved or weakened test performance.
Technical approach
- Applied feature engineering to structure the input representation.
- Compared decision trees, KNN, and neural networks.
- Analyzed overfitting and test-set behavior.
- Interpreted model tradeoffs rather than treating one score as the whole result.
What I built / contributed
I built the classifier workflow, compared model behavior, and used the evaluation results to reason about generalization rather than selecting a model from one headline score.
Result or evaluation
The project reinforced a workflow I continue to value: define useful features, establish baselines, compare consistently, and interpret where performance comes from.
Tools
Python, model comparison, feature engineering.
Links and availability
This is a course project. Public artifacts should be attached only after confirming the data and code can be shared.