Grouping Naive Bayes model based multi-factor online prediction method and system

The present invention discloses a grouping Naive Bayes model based multi-factor online prediction method and system. Multiple factors are grouped in a classification or clustering manner, so as to enable factor groups to be independent from each other but the interior of each factor group to be highly correlated to each other, so that effective model learning can be performed by using a Naive Bayes model, which makes factor prediction meet requirements of the Naive Bayes model and thus implement multi-factor learning in the case that a training data set is limited. Moreover, by introducing a dynamic discretion upgrade method, the calculation amount and time complexity of online learning are greatly reduced, online real-time learning and prediction are implemented, and the method and system disclosed by the present invention can be used in a complex system in which model feature data changes relatively fast by the time to perform multi-factor online prediction, and in particular, be suitable for online predict