The invention relates to data mining, and discloses a data mining method and device. The method comprises the steps of collecting data of event sets with complete information and data of event sets with incomplete information, utilizing the data of the event sets with the complete information for training at least one prediction model, obtaining the relevancy of the data of the event sets with the incomplete information and the prediction models, enabling the data sets with the incomplete information to be matched with the corresponding prediction models according to the obtained relevancy, and utilizing the data of the prediction models matched with the event sets with the incomplete information for supplementing the data of the event sets with the incomplete information. According to the data mining method, dynamic data of events and the data of the prediction models are utilized for supplementing lost data, and precise data supplementation can be achieved even when known static data are few or poor in quality.