Abstract
Data classification came into widespread in various fields in recent years. However, basically no one paid attention to the discount prediction on the biggest game website in the world-Steam. In this work, we aim to predict whether the game on Steam is on discount or not. First, we collect data from the official website and model them via machine learning based models, including Logistic Regression and Random Forest, test their performances via the true data, and we conclude that the Random Forest achieves the best performance, which reaches the accuracy of 79.5%. This model will help gamers to save money while making purchases. Moreover, we can predict the popularity of game in our future work.