Book recommender system

Question

I have a user who has bought 5 books of different genre ie fiction, fiction, fiction, n-fiction, self-help. I have this data in a dataframe.

UserID  AGEGROUP    GENDER                    genre 
600003     0          1      [fiction, fiction, fiction, n-fiction, self-help]  

Similarly, I have data for 15000 users including data like type of books bought(paper white, e books etc), genre. Using this data, I need to predict books a user is likely to buy from a different data set(different books but same variables(Book type, genre)).

My question is- for every user, how do I calculate the likelihood of purchasing a book. Do I take the mode of genres, book type etc(In the above DataFrame, I will have on Fiction as it is repeated for most number of times instead of having 5 genres) for a user and try to find the nearest book in the new data set? Or is there any other approach I can take to predict genre and book type for every user.
PS: I am new to data science and python

Answer 1

Collaborative filtering may be the exact thing what you are looking for. It simply does:

User A buys book of type {1,2,3}

User B buys book of type {2,3,4}

So the system suggests type 4 to User A and type 1 to User B.

It is based on user similarity rather than book similarity. You can find detailed information here .