Research on a Game Data Prediction Method Based on an Improved GRU Algorithm

With the rapid development of the gaming industry, massive amounts of player behaviour data provide a wealth of reference for operational decisions. However, traditional statistical methods are limited in their effectiveness when dealing with complex, high-dimensional, non-linear gaming data. To this end, this paper proposes a game data prediction method based on an improved GRU algorithm. By introducing the Attention mechanism, the model can better capture the long-term dependencies in game data and combined with an adaptive parameter adjustment module to improve training efficiency and model stability. This paper discusses data pre-processing in detail, model architecture design, parameter optimisation, etc., and experimentally demonstrates the significant improvement in prediction accuracy and stability of the improved GRU model. The experimental results show that the improved GRU model not only outperforms the traditional GRU on multiple evaluation metrics, but also exhibits greater robustness and practical applicability.