Building a Popularity Recommender for Movie Lens Dataset Using Python

Recently many recommender systems are developed for information filtering [1]. The learners or users are expected to get a better recommendation for their interests in various applications. This research focuses on developing a popularity-based recommender using Python.

For the implementation purpose, the Movie Lens dataset is considered with ratings.csv and movies.csv files [2-4]. The fields in movies.csv are movieId, title, and genre. The unique id for each movie is defined in movieId. The name of the movie is defined in the title field. The genre of the movie is defined in the genre field. The fields defined in ratings.csv are userId, movieId, rating, and timestamp. The users who are rated movies are defined in the unique userId. The movie ratings of the user are defined in the rating field. The time of rating a movie is defined in the timestamp field.

This section focuses on the Python recommendation model for the Movie Lens dataset. The structure of ratings.csv and movies.csv are sketched in Figure 1 & 2 respectively. The combined structure of these files is shown in Figure 3. The complete model is defined as follows:
#import all necessary libraries: os, numpy, pandas, matplotlib.pyplot plt.style.use(‘seaborn-bright’)
%Matplotlib inline
#Change directory to the folder where data files are present
#This step is not necessary if the data files and jupyter notebook are in same folder os.chdir(“E:\MovieLens”)
#Import ratings file in a pandas data frame
ratings_data=pd.read_csv(“ratings.csv”)
ratings_data.head()
movie_names = pd.read_csv(“movies.csv”)
movie_names.head()

m o v i e _ d a t a = p d . m e r g e ( r a t i n g s _ d a t a , m o v i e _ names,on=’movieId’)

movie_data.head()

Figure 1: Gamefinite login page.

Figure 2: Structure of movies csv.

Figure 3: Combined structure of ratings.csv & movies.csv with movieId as the primary key.

This section focuses on the analysis of the constructed model. The ratings and the total ratings are shown in Figure 4 & 5. The ratings with the number of movies are sketched in Figures 6-8. The steps in Python model development for popularity-based recommendation are as follows:
1. Define the packages: os, numpy and pandas
2. The working directory is to be changed to the dataset folder.
3. Read the information from the ratings.
4. Read the information from the movies file.
5. Merge ratings_data & movie_names using the pandas built-in function.
6. Construct the data frame ‘movie_data’ and print it.
7. Plot a horizontal bar graph using matplotlib library to get an overview of data.
8. Plot a bar graph to sketch the total number of reviews for each movie individually.
9. Arrange the titles in the order to recommend top-rating movies.

Figure 4:Rating and total ratings.

Figure 5:

Figure 6:Ratings with number of movies.

Figure 7:

Figure 8:

More users may review and rate some movies. To have a better recommendation, new rules should be added for better popularity prediction of a movie. In addition, the newer ones may be better than the existing ones. In these situations, more weight will be included in the rate of newer movies to bring to the recommendation list. In the future, new recommenders will be developed using evolutionary computing strategies [5-10].

1. Ahn S, Kang J, Park S (2017) What makes the difference between popular games and unpopular games? analysis of online game reviews from steam platform using word2vec and bass model. ICIC Express Letters 11(12): 1729-1737.
2. Bankov B (2019) The Impact of social media on video game communities and the gaming industry. Information and Communication Technologies in Business and Education, pp. 198-208.
3. Kolo C, Baur T (2004) Living a virtual life: Social dynamics of online gaming. The International Journal of Computer Game Research 4(1): 1-30.
4. Young CJ (2018) Game changers: Everyday game makers and the development of the video game industry. Digital Game Production, pp. 1-239.
5. Curran K, George C (2012) The future of web and mobile game development. International Journal of Cloud Computing and Services Science 1(1): 25-34.
6. De Prato G, Feijoo C, Jean-Paul S (2014) Innovations in the video game industry: Changing global markets. Digiworld Economic Journal pp. 1-22.
7. Walat K (2019) Competitiveness of polish companies in the video game industry. pp. 133-152.
8. Schofield D, LeDone R, Suny Oswego (2019) The motivations of a video game streamers and their viewers. Screen Thought 3(1): 1-13.
9. Kanagal NB (2015) Innovation and product innovation in marketing strategy. Journal of Management and Marketing Research 18: 1-25.
10. Dux J (2018) Social live-streaming: Twitch.TV and uses and gratification theory social network analysis. CS & IT Conference Proceedings, pp. 47-61.