Recommendation systems are an important application of data science. They provide recommendations of items to sell to users on Amazon and movies/shows to watch on streaming platforms such as Netflix. The most applied approaches to recommendation based systems are Collaborative filtering and Content based filtering. Collaborative-based filtering is a method of recommending products to customers using their past behaviors or ratings as well as similar decisions by other customers to predict which items might be appealing to the original customers. We utilize the memory-based approach of collaborative-based filtering Content-based filtering suggests products to customers by using the characteristics of an item in order to recommend additional items with similar properties. Collaborative filtering is also more popular for recommendations where there is a limited amount of data by each user or item.
Once the window opens, you are presented to an introduction screen. As advised, please select one of the two algorithms to get movie recommendations based off of the selected recommender system algorithm.
To get recommendations based on Collaborative-based filtering: Select Collaborative-Based filtering to test collaborative based filtering. You are then prompted to click 3 boxes of genres you are interested in getting movie recommendations from. Click next to continue. You then open a new window called “Movies”. The window has the top 2 movies from each genre listed on the left. On the right, there is an entry box and you are prompted to enter your favorite movie. Enter your favorite movie then select next to continue to the next screen to get recommendations. The final window displays recommended movies based on your interests. Exit to return to the first window with menu options.
To get recommendations based off Content based filtering: Select Content-Based filtering to test content-based filtering. You are then prompted to enter a favorite movie in the entry box. Press enter to get recommendations. The window then displays recommended movies based off of your interests. Exit to return to the first window with menu options.
The collaborative-based filtering data we will use comes from MovieLens and can be found at https://grouplens.org/datasets/movielens/ . We chose the “small” dataset of 100,000 ratings under the subheading “recommended for education and development”. Contains multiple CSV files that include: (“Recommended for education and development; Small”) 100,00 ratings of 9,000 movies made by 610 users, and 3,600 tag applications. Users did not rate every single movie listed. A list of these movies with associated genres A list of these movies with keywords/tags
The content-based filtering data we will use comes from Kaggle and can be found at https://www.kaggle.com/rounakbanik/the-movies-dataset.