Skip to content

Python library for implementing Content-Based Recommendation Engines with ease!

License

Notifications You must be signed in to change notification settings

mochatek/cbrecommender

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

14 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

cbrecommender

Downloads Downloads Downloads

cbrecommender is a Python library for implementing Content-Based Recommendation Engines with ease!

A Content-Based Recommender is a form of Personalized recommendation System that maintains a user profile and tries to match the items with the taste profile of a user before presenting them as a recommendation to the user.

The key ideas are:

  • Model items according to relevant attributes derived from the content.
  • Develop a user profile either from their implicit actions (clicks, time spend on a video etc.), explicit actions(purchase, rating etc.) or by combining both.
  • Use these profiles to provide recommendations.

Installation

Install from pypi with pip :

pip install cbrecommender

Usage

1. Importing and initializing :

from cbrecommender import CBRecommender

recommender = CBRecommender()

2. Creating Item Profiles :

In Content-Based Recommender, we must build a profile for each item, which will represent the important characteristics of that item.

item_profiles = recommender.create_item_profile(features)
  • features: DataFrame must be relevant attributes of the item that signifies the user's preferences. For example: movie genres, news topics, post tags etc.

  • create_item_profile() -> DataFrame will return the item_profiles created from the supplied features.

3. Creating User Profile :

user_profile = recommender.fit(train_item_profiles, scores)
  • fit() -> DataFrame is where we extract user preferences from the item-profiles and associated scores, and then construct the user-profile.

  • train_item_profiles: DataFrame must be a subset of the item-profiles created at step 2. For example, it can be the item-profiles of the movies already watched by the user (watch history).

  • scores: List[float] must be a list of some measure corresponding to each item in train_item_profiles, denoting how much the user liked that item. For example: Rating for a watched movie, song etc.

4. Get recommendations based on User Profile :

recommendations = recommender.recommend(test_items, test_item_profiles, min_score, limit)
  • test_items: DataFrame must be those items that the user have not used for training and from which we need recommendations. For example: Unwatched movies.

  • test_item_profiles: DataFrame must be the item-profiles of test_items.

  • min_score: float must be a numerical value (1-10) that specifies the minimum score for recommending items. Default is 7.5.

  • limit: int must be an integer that denotes the number of items to recommended.

Example

from cbrecommender import CBRecommender
from pandas import DataFrame
data = DataFrame(
{'movie':['Endgame','Avatar','Titanic','Infinity War','Jurassic World','Black Panther',
          'Harry Potter-II','The Last Jedi'],
 'genre':['Action,Adventure,Drama','Action,Adventure,Fantasy','Drama,Romance',
          'Action,Adventure,Sci-Fi','Action,Adventure,Sci-Fi','Action,Adventure,Sci-Fi',
          'Adventure,Drama,Fantasy','Action,Adventure,Fantasy']
})
print(data)
movie genre
Endgame Action,Adventure,Drama
Avatar Action,Adventure,Fantasy
Titanic Drama,Romance
Infinity War Action,Adventure,Sci-Fi
Jurassic World Action,Adventure,Sci-Fi
Black Panther Action,Adventure,Sci-Fi
Harry Potter-II Adventure,Drama,Fantasy
The Last Jedi Action,Adventure,Fantasy
recommender = CBRecommender()

# We are considering genre alone as the feature. You can include other features as well.
movie_profiles = recommender.create_item_profile(data[['genre']])
print(movie_profiles)
action adventure drama fantasy romance sci-fi
1 1 1 0 0 0
1 1 0 1 0 0
0 0 1 0 1 0
1 1 0 0 0 1
1 1 0 0 0 1
1 1 0 0 0 1
0 1 1 1 0 0
1 1 0 1 0 0
# Consider we had watched the first 4 movies. So we use it as training data to extract preferences.
# We use the user rating for the watched movies as the preference score.
watched_movie_profiles = movie_profiles.iloc[:4, :]
watched_movie_ratings = [8.5,7.8,7.8,8.5]

user_profile = recommender.fit(watched_movie_profiles, watched_movie_ratings)
print(recommender.user_profile)
action adventure drama fantasy romance sci-fi
0.2755 0.2755 0.1811 0.0866 0.0866 0.0944
# We use the remaining 4 unwatched movies as test data to get recommendations from.
unwatched_movies = data[['movie']].iloc[4:,:]
unwatched_movie_profiles = movie_profiles.iloc[4:,:]

# Recommend top 3 movies with minimum expected rating of 5.0
recommendations = recommender.recommend(unwatched_movies, unwatched_movie_profiles, 5.0, 3)
print(recommendations)
item expected score
Jurassic World 6.45
Black Panther 6.45
The Last Jedi 6.37

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

License

MIT License

About

Python library for implementing Content-Based Recommendation Engines with ease!

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages