Research on Influencing Factors of Video Game Sales using Binary Logistic Regression

Abstract

In recent years the game industry welcomed rapid development. As a result, it will be meaningful to figure out what factors influence the sales of games and how they influence them. Based on games from Electronic Arts, this paper intends to use binary logistic regression to analyze the effects of platform, genre, critic score, user score, and rating. The dependent variable is sales in North America. The data is from Kaggle, Metacritic, and the Entertainment Software Ratings Board. According to the regression model, which has an acceptable predicting accuracy of 80.09%, it can be inferred that critic score, platform, and user score represent the significance, while genre and rating don’t show a significant relation with sales. Among the three significant variables, critic score has a positive correlation with sales, while user score has a negative correlation. This means games with high sales usually have high critic scores but do not always have high user scores. As a categorical variable, the platform's significance means different platforms may suit different games in general. By contrast, players do not show a preference for a certain game genre or some intense content.