This algorithm recommends games based on user Steam library.
Pandas and numpy for data modification.
Collaborative Filtering based on Item similarity:
- Cosine_similarity from sklearn
Collaborative Filtering based on User similarity:
- Pearsonr algorithm from scipy.stats
For gathering data from Steam
- Many thanks to zeo's steam project
To run this project, paste Your steam Id and steam API key here and run steamUserData.py file.
- steam API Key
- steam Id (It's a 40s video, this wasn't created by me)
PS The datasets that I've created are to large to post on github, If you want to avoid training data yourself, write me a message and I'll give You GoogleDisc link with finished dataframe.
PSS. If I had to write that project again I would use other file format like parquet or feather since that dataset would then be much smaller and I don't need to visualise that data.