In this repository you will find my project for the Network Science exam at UniMi.
In this case, I took a dataset from Kaggle containing 41 million game-reviews published on the Steam platform from October 2010 to December 2022.
The objective was to create a Recommendation System for popular Steam games using reviews from June 2022 to December 2022 using the unsupervised learning algorithm Node2Vec.
I first preprocessed the afore mentioned dataset using R, in particular with the dplyr package of the tidyverse. Then I moved to Python where I created the graph thanks to the networkx library which was also the tool of choice to perform the study.
The obtained graph, already color-coded according to the Louvain communities is:
Overall, Node2Vec performed well suggesting games of the same saga or of the same genre.
If you have any questions or suggestions feel free to contact me via e-mail or on linked-in! 💻🎮