Online Mini-Game Popularity and Feedback Data Prediction Based on Time Series and BP Neural Network Prediction

In recent years, in order to increase network traffic, online mini-games have accounted for an increasing proportion of the video game market share. For example, the New York Times acquired the online mini-game “Wordle” in order to attract customers, but the online mini-game has the problem of large fluctuations in popularity and large changes in player feedback, and the current research on related issues at home and abroad is still insufficient, resulting in online mini-game companies unable to accurately judge the popularity of the game and player experience and make wrong judgments. This has led to a more serious customer churn. To this end, this article uses the New York Times mini-game “Wordle” player data from January 2022 to January 2023 as an example. An exponential smoothing prediction model and a BP (Back Propagation) neural network prediction model are established to predict the number of players and scores of Wordle, respectively, to realize the prediction of online game popularity and player feedback. The overall fitting error of the final population prediction model does not exceed 3,000 people, and the MAE (Mean Absolute Error) of the fitting error of each of the seven outputs of the score prediction model does not exceed 5. Experimental results show that the two models used in this paper have high feasibility for the prediction of popularity and player feedback of online mini-games.