Pick a Kit
Meal kit comparison platform — browse and filter thousands of meals across services by protein, price, and ratings. Includes a machine-learning protein classifier and a survey-driven recommendation engine.
Pick a Kit started as a question: which meal kit service is actually worth it? I built a web scraper to collect meals data from a dozen services, then a comparison site so users could browse and filter by protein, cuisine, price, and ratings.
The protein classifier — trained with a random forest on tokenized meal names and descriptions — automatically tags thousands of meals as beef, chicken, seafood, vegetarian, and so on, so users can instantly filter without manual tagging.
A later survey collected real user ratings, enabling item-based collaborative filtering recommendations. The project became the basis for my master's thesis in data science at CSULB.
Screenshots
