Skip to content

The repository consists of a recommendation engine that suggests movies to the users based on the genre and ratings previously received. Under the hood, a neural collaborative filtering technique has been implemented

License

Notifications You must be signed in to change notification settings

Tejas-TA/Neural-Network-Movie-Recommendation

Folders and files

NameName
Last commit message
Last commit date

Latest commit

959c0ab · Aug 1, 2021

History

7 Commits
Jun 10, 2021
Jun 10, 2021
Aug 1, 2021

Repository files navigation

Neural Network based Movie Recommendation

Project Walkthrough

Movie recommendation is based on the neural collaborative filtering technique. Movielens dataset consisting of 100K records is analyzed.

Phase 1 - Data preprocessing

  1. Data Analysis and manipulation using pandas
  2. Data Visualization using matplotlib
  3. Test Train split

Phase 2 - Neural Network building

  1. Created movie embedding path
  2. Created user embedding path
  3. Concatenated movie and user vectors
  4. Weight initialization using the random uniform technique
  5. Added 5 fully connected layers with activation function as ReLu
  6. Adam optimizer used
  7. Model has complied

Phase 3 - Training

  1. Model is run for 18 epochs with verbosity 1
  2. Achieved loss of 0.6
  3. As seen below graph of Training error vs epoch, there is a reduction in error continuously after epoch=5

Screenshot 2021-07-18 at 10 28 31 PM

  1. Minimum RSE of 0.8 is at epoch 18

Phase 4 - Movies suggested to the user

Screenshot 2021-07-18 at 10 35 10 PM


Email - tejasta@gmail.com
LinkedIn - https://www.linkedin.com/in/tejas-ta/
Blogs - https://tejasta.medium.com/