write code in matlab to build an svd recommendation system in matlab

main.m
ratings = importdata('ratings.csv'); % Load your rating data file
66 chars
2 lines

main.m
% Normalize the ratings
normalized_ratings = ratings - mean(ratings, 'all');

% Create the user-item matrix
user_item_matrix = sparse(normalized_ratings(:, 1), normalized_ratings(:, 2), normalized_ratings(:, 3));

% Perform SVD
[U, S, V] = svds(user_item_matrix, k); % k is the number of singular values to keep
312 chars
9 lines

main.m
predicted_ratings = U * S * V'; % Reconstruct the user-item matrix using the truncated SVD

% If you want to predict the rating of user u for item i:
u = 1; % user index
i = 1; % item index
predicted_rating = predicted_ratings(u, i);
234 chars
7 lines

main.m
% Split your data into training and testing sets
train_ratio = 0.8; % e.g., 80% training, 20% testing
[train, test] = split_data(ratings, train_ratio);

% Calculate the mean squared error (MSE) to evaluate the recommendation system
test_user_item_matrix = sparse(test(:, 1), test(:, 2), test(:, 3));
predicted_test_ratings = U * S * V';
mse = mean((predicted_test_ratings - test_user_item_matrix).^2, 'all');
409 chars
9 lines