Recommendation systems



  • Hi everyone, last weekend I built a model for recommendation systems over MovieLens 100K Dataset. There are two kinds of system,

    1. The collaborative algorithm uses user behavior for recommending items. They use behavior of other users in terms of usage history, ratings, selection and purchase information. Other users behavior for the items is used to recommend items to the new users. In this case, features of the items are not known.

    2. The content-based algorithm is that we have to know the content of both user and item. From data, we construct user-profile and item-profile using the content of shared attribute space. In our example, for a movie, you represent it with the movie genres (as a multi-hot vector). For user profile, you can do the same thing based on the users likes some movie stars/genres etc.

    Examples:
    a. Collaborative based recommendation system based on library LightFM. https://beta.datmo.io/shabazp/recommendation-system
    b. Content based recommendation based on library Surprise. https://beta.datmo.io/shabazp/content-based-recommendation-system

    Please do share your thoughts.



  • @shabazbpatel there's not really much information about your collaborative recommendation system (https://beta.datmo.io/shabazp/recommendation-system) in the README. Could you share more ? I am not sure how I can use it with my own user data.

    Can you give an example either in the README of the type of data?



  • HI @anand, I have added more information about the model. Hope that helps. :)


Log in to reply
 

Looks like your connection to Datmo was lost, please wait while we try to reconnect.