Sample Movie 📽️ Recommendation System 🔢 using Cosine Similarity ⚙️

cosineSimilarityRec.py

import numpy as np
from sklearn.metrics.pairwise import cosine_similarity

# Sample ratings matrix for 5 movies by 5 users
ratings = np.array([
    # [Movie 1, Movie 2, Movie 3, Movie 4, Movie 5]
    [5, 3, 4, 4, 0],  # User 1
    [3, 1, 2, 3, 3],  # User 2
    [4, 3, 4, 3, 5],  # User 3
    [3, 3, 1, 5, 4],  # User 4
    [1, 5, 5, 2, 1]   # User 5
])

# Calculate cosine similarity between users
similarities = cosine_similarity(ratings)

# Rating prediction function for a user and item
def predict_rating(user_index, item_index):
    # Initialize variables for prediction
    weighted_sum = 0
    similarity_sum = 0
    # Iterate over all users
    for i in range(len(ratings)):
        # Exclude the user itself and users who have not rated the item
        if i!= user_index and ratings[i][item_index]!= 0:
            # Find the set of items rated by both users
            common_items = np.nonzero(ratings[user_index] * ratings[i])[0]
            # Adjust similarity and calculate weighted sum for common items
            if len(common_items) > 0:
                adjusted_similarity = similarities[user_index][i]
                weighted_sum += ratings[i][item_index] * adjusted_similarity
                similarity_sum += adjusted_similarity
    # Predict the rating using weighted average
    if similarity_sum!= 0:
        predicted_rating = weighted_sum / similarity_sum
    else:
        predicted_rating = 0
    return predicted_rating
  
# Predict rating in sample matrix for user index 0, item index 4 
predicted_rating = predict_rating(0, 4)
print("Predicted rating for User 1 on Movie 1:", round(predicted_rating,2))