|（Department of Information Technology and Engineering，Guangzhou Business College，Guangzhou 510363，China)
Abstract: Based on big data, the authors carried out classification and prediction on university student fitness, which is conducive to university sports governance system construction; the Naïve Bayes model is a machine learn-ing classification algorithm that is simple to operate and provided with good performance. Based on Naive Bayes classification algorithm, and using the physical test data of classes 2014 and 2015 students of Guangzhou Business College and their score results as source data, the authors established a university student fitness classifier. By ap-plying such a classifier, researchers can, in a certain sense of probability, correctly determine newly or previously enrolled university students’ fitness condition, thus give a proactive early warning to those students whose fitness has a relatively high probability of hidden troubles, so that university physical education can carry out group fitness determination and individualized effective intervention on the students, thus promoting student healthy development and improving university students’ overall fitness level. The classifier mode was realized by using Python coding, in the end, the classifier’s accuracy rate was verified by using historical fitness data that did not overlap with training data, and the result showed that the fitness classifier based on naïve Bayes algorithm reached a correct rate of 78%.
Key words: school physical education；university student fitness analysis；sports intervention；Naive Bayes classifier algorithm；big data