search

NicolasHug / Surprise

A Python scikit for building and analyzing recommender systems
http://surpriselib.com
BSD 3-Clause "New" or "Revised" License
6.42k stars 1.02k forks source link

Cross-Domain Filtering #196

Open mohdkashif93 opened 6 years ago

mohdkashif93 commented 6 years ago

Was I hoping to implement a cross-domain filtering module in this package? I have already done it before as part of my college project and was able to achieve almost similar accuracy. I was hoping to start with this

Can movies and books collaborate ?

Then add more methods over time. Please let me know your thoughts.

NicolasHug commented 6 years ago

Thanks for the proposition, it can be interesting. How do you see the different input methods changing, as well as the cross-validation pipelines?

As always with the addition of a new feature, the main issue is to make sure that it can integrate well with the current tools, and it's not always easy or obvious.

mohdkashif93 commented 6 years ago

The only difference that I can think of right now is that two datasets will be supplied to the all the functions instead of one. Rest everything remains the same. I am thinking of implementing it separately from your package first, and then integrating it once you have a look at it and see what design changes you would like?

NicolasHug commented 6 years ago

The only difference that I can think of right now is that two datasets will be supplied to the all the functions instead of one.

I agree but that's really not a trivial change ;)

Yes doing it separately for now I'd a good idea. I hope you understand however that I cannot guarantee that we will be able to integrate the changes eventually. This will depend on how well they fit with the current codebase.

mohdkashif93 commented 6 years ago

Yes, I understand. :-)

On Tue, Jul 31, 2018 at 6:59 AM, Nicolas Hug notifications@github.com wrote:

The only difference that I can think of right now is that two datasets will be supplied to the all the functions instead of one.

I agree but that's really not a trivial change ;)

Yes doing it separately for now I'd a good idea. I hope you understand however that I cannot guarantee that we will be able to integrate the changes eventually. This will depend on how well they fit with the current codebase.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/NicolasHug/Surprise/issues/196#issuecomment-409064053, or mute the thread https://github.com/notifications/unsubscribe-auth/AUHDSsB8G1GRaVNm9s8EB7PJ30SjCPocks5uL7L9gaJpZM4VljZW .