A random forest algorithm based customer demand forecasting model for sports enterprises in the real economy

The study proposes two strategies to optimise the random forest algorithm (RF) by fine-tuning the data distribution and introducing customer life values, constructing the improved random forest algorithm (IRF), and building a customer churn prediction model based on the IRF algorithm. A churn segmentation model is constructed based on the k-means algorithm to classify customers according to their characteristics in order to predict their needs and thus develop differentiated strategies to retain them. The experimental results show that the IRF prediction model has an accuracy of 99.84% and an AUC value of 0.932. The above results show that the accuracy of the IRF algorithm can meet the actual demand, which will contribute to the long-term development of sports enterprises. In the future, it is necessary to consider the relationship and interaction between customers to further improve the prediction accuracy of customer demand.