Back to projects
Machine Learning

Food Recommendation System

Used KNN to build a simple food recommendation engine based on user preferences.

  • Python
  • KNN
  • Machine Learning

Preference-based recommendation

A compact recommender built for the Pathao Challenge 2019 food-delivery dataset (user, item, restaurant, cuisine, and order-time features). Instead of a content model per restaurant, it recommends by proximity in preference space: encode each user's order history into a feature vector, find the users whose vectors are closest, and surface the items they liked that the target user hasn't ordered yet.

User preferencesorder/rating historyNearest neighborssimilar usersRecommendationsunseen items, ranked

Pipeline

  • Categorical fields (cuisine, category, restaurant) label-encoded into a numeric feature space alongside order-time features (day of week, hour of day, item count).
  • Distance-based neighbor lookup over that feature space to group users with similar ordering behavior, then rank candidate items by how often a user's neighbors ordered them.
  • Deliberately simple. A baseline collaborative-filtering approach rather than a learned ranking model, scoped to what a small labeled dataset supports.