Recommendation-System-based-on-Nonnegative-Matrix-Factorization

Goal:

To recommend the movies that user might want to watch.

Recommender systems

• Recommender systems aim to predict user’s interests and recommend product items that quite likely are interesting for them. They are among the most powerful machine learning systems that online retailers implement in order to drive sales.
• Data required for recommender systems stems from explicit user ratings after watching a movie or listening to a song, from implicit search engine queries and purchase histories, or from other knowledge about the users/items themselves.
  
  Examples of Recommendation systems are Netflix or YouTube that suggest playlists or make video recommendations

Types of recommender systems:

• Content-based systems, which use characteristic information.

• Collaborative filtering systems, which are based on user-item interactions.

• Hybrid systems, which combine both types of information with the aim of avoiding problems that are generated when working with just one kind.

Dataset :

Movie lens Dataset consists of (link):

• 100000 ratings (~ 1 lakh)
• 600 users

Some Analysis :

Information about the data :

---

Number of Users : 610
Number of movies : 9742

---

Movies with highest number of user ratings :

Age of Innocence, The (1993)
I Love Trouble (1994)
Virtuosity (1995)
Cemetery Man (Dellamorte Dellamore) (1994)
Teenage Mutant Ninja Turtles II: The Secret of the Ooze (1991)

---

User who gave more ratings :

UserId	Movie ID
414	2698
599	2478
474	2108
448	1864
274	1346

---

Matrix factorization

Where,

• R (users,movies)
• U (users,d)
• VT (d,movies) Here, d is the number of latent features.

Cost Function

Prediction of rating of movies using gradient descent algorithm.

Gradient descent

Gradient descent is a first-order iterative optimization algorithm for finding a local minimum of a differentiable function. To find a local minimum of a function using gradient descent, we take steps proportional to the negative of the gradient (or approximate gradient) of the function at the current point.

Optimization techniques used in Gradient descent

• Regularized Gradient descent

• Sliding Window Gradient descent

• Line Search Gradient descent

• Particle Swarm Optimization(PSO) Gradient descent

Results :

Root Mean Square Error (RMSE)

Optimization	Epochs	Train error	Test error
Gradient Descent (MMF)	10000	1.09	1.14
Regularised (MMF)	1000	1.43	1.46

It's time to recommend :

For User Id : 2

Movie ID	Movie
26776	Porco Rosso (Crimson Pig) (Kurenai no buta) (1992)
104879	Prisoners (2013)
5358	Mountains of the Moon (1990)
175569	Wind River (2017)
626	Thin Line Between Love and Hate, A (1996)