A Restaurant and Dish Recommender system using user profile, order history and ratings based similarity computation.
The users are presented with a choice of three recommendation models. The first one is a content-based filtering model, whereas the second and third are keyword-based filtering models. In the third model, recommendations are modified based on user feedback.
Frontend: Javascript, HTML, CSS
Backend: Python , Django , MySQL / Postgres SQL
others scikit-learn , numpy , pandas
-
Utilizing person preference to recommend more personalized food items.
-
Content based recommendation system recommends items based on the content of items. I.e. features of items
-
First model is based on using TF-IDF vectorization and cosine similarity.
-
Second model is based on the keyword extracted from dishes like ingredients, flavor, profile etc.
-
Third model is based on penalization of keywords and promotion and demotion of certain keywords by feature vectors of the user based on the feedback received from users.