Research on Abnormal Identification of User Behavior Data on Sports Mobile Education Platform

To address the problems of high error rate in feature extraction, low abnormal recognition rate, and long recognition time in traditional methods for sports mobile education platform user behavior data, a new method for anomaly detection of user behavior data on sports mobile education platforms is proposed. This method involves denoising the user behavior data on sports mobile education platforms, extracting features from the denoised data using the STL algorithm, and selecting data features using the MRMR algorithm. By combining the feature selection results with the decision tree algorithm, the dataset is divided into normal and abnormal subsets, thereby achieving anomaly detection of user behavior data on sports mobile education platforms. Experimental results show that the error rate of feature extraction in this method varies between 3% and 5%, the abnormal recognition rate varies between 94% and 98%, and the average recognition time is 0.57s, indicating a high recognition accuracy.