Skyrim Game Mods Endorsement Prediction with Machine Learning

Gardyan Akbar; Vincentius Tandra; Nunung Nurul Qomariyah

Mods within video games have become an important part of the gaming experience for many, with the activity spawning its own dedicated community of developers and players. We decided to explore one of our questions. What makes a mod popular to others? For our research, we decided to analyze the attributes of the mods on the major PC game title The Elder Scrolls V: Skyrim Special edition. Using natural language processing techniques, we attempted to find out which features affected the popularity of the mod and we did so through the measure of endorsement counts that a mod had. During early research, we found a correlation between the inclusion of adult content within mods and the effect it had on the final endorsement count of the mod. We were able to find out that adult content was indeed an important factor in determining the endorsement count of a mod and hope to answer more questions about the popularity of mods through the analysis of demos and pictures. We use ensemble machine learning model in this study. Our experiment shows that the model’s accuracy in predicting the endorsement count of a mod was also very high, yielding a score of 93%. We hope that our research is able to assist the future work of mod and game developers alike.