search

GeekHaven / GeektoberFest-AI

This repository contains the basic projects related to Data Sciences and AI.
2 stars 8 forks source link

Movie Recommender System : Apply Sparse SVD and compute RMSE #31

Closed TamoghnoBhattacharya closed 4 years ago

TamoghnoBhattacharya commented 4 years ago

We'll start with our collaborative filtering model. This is based on matrix factorization using singular value decomposition (Yes, the one from Linear Algebra). Follow the given procedure -

  1. Use the code created in issue #3 to create user rating matrix. You will find it under Movie Recommender System/IIT2019134/rating_matrix.py.
  2. Now apply Sparse SVD available in scipy, to the matrix to get the 3 matrices - U, sigma and V-transpose. Choose the latent factors argument wisely to avoid over-fitting and under-fitting. I recommend using in the range 40-60.
  3. Multiply the 3 matrices back together to generate the matrix of predicted ratings.
  4. Compute the RMSE, i.e. Root Mean Squared Error between the actual ratings matrix and the predicted ratings matrix, and print it. Note that RMSE is only computed on the values which were not NaN in the original ratings matrix, so be careful.
JLodha commented 4 years ago

I want to work on this

Lazy-Leopard commented 4 years ago

@JLodha Go on.

aarpit1010 commented 4 years ago

@Lazy-Leopard @TamoghnoBhattacharya Actually, I have completed the work on this issue, can I submit the PR?

TamoghnoBhattacharya commented 4 years ago

@Lazy-Leopard @TamoghnoBhattacharya Actually, I have completed the work on this issue, can I submit the PR?

Consult with @JLodha and decide who's going to make the PR first. The first correct PR will be merged, simple as that.

JLodha commented 4 years ago

Since I claimed the issue first, though there is no hard and fast rule. I would like to request the admins/collaborator to give me a day more (as per the 2 day policy). Hoping a positive response. Thanks

Lazy-Leopard commented 4 years ago

@JLodha Go on,complete it by today.