This paper analyzes and studies the tactics in basketball games using the main branches of cluster analysis using multivariate statistical methods. A brief overview of multivariate statistical analysis is summarized by analyzing the application of multivariate statistical methods. The main method for classifying basketball tactics is cluster analysis, and nonlinear programming functions with constraints are constructed using mean square approximation theory. The classification results are reasonable when combined with the least square error sum. According to the results, man-to-man defensive tactics were most frequently used in basketball, with 72 instances accounting for 75.19% of all defensive strategies. The most frequent offensive tactic in the basketball game was dribbling with a fast break, and 15 times succeeded in scoring during 20 dribbling with a fast break in the first half, with a success rate of 75%.