The application of GRU (gated recurrent unit) model in the correlation analysis of training load and skill improvement in table tennis teaching under the internet of things environment

The connection between the training load and table-tennis skill development has become increasingly significant with the rise of smart IoT technologies in sports education. Traditional assessment methods fail to capture temporal dependencies and dynamic performance variations, limiting their effectiveness for real-time training adaptation. Research applies deep learning (DL) techniques to model and predict the relationship between training load and skill progression using an IoT-integrated framework. Data were sourced from a Kaggle dataset containing structured training load, performance metrics, and skill-score measures. Pre-processing included data cleaning, normalization, and handling of missing values. Time-Domain Feature Extraction (TDFE) is applied to derive stroke dynamics, fatigue indicators, and temporal sequences required for skill assessment. The GRU model successfully captured the non-linear temporal interactions between intensity, duration, stroke frequency and skill output variables, and the MOA optimized the IoT feature selection and optimized the GRU model parameters for better prediction. The parameter influence was evaluated using the Pearson Correlation Analysis (r) between the framework outputs and actual skill measures. The experimental results showed that the proposed MOA+GRU model has an accuracy of 96.65% and an AUC of 0.97, which are better than the existing models. Python is used to implement the experimental setup, and the integration of GRU modeling and MOA optimization for feature selection. The combination of MOA and GRU provides a scientifically solid, real-time training system for coaches, allowing them to provide personalized feedback, monitor training more effectively and develop table-tennis skills more effectively.