Wherein for analyzing the causal effect extraction method and apparatus

The present invention discloses a feature extraction method and apparatus for analyzing the causal effects, belonging to the field of data analysis. The method comprising: determining a characteristic point in time of the event analysis of the results of the causal effect; according to the characteristic point in time, acquire a preset number of time intervals, said predetermined number of time intervals in said feature point in time before , the time interval and distance interval of the span length of the time interval feature time point positive correlation; the potential causes of the event information for each time interval the event occurred, event extraction results for the analysis of causal effects features. In the case of the present invention is considering the potential causes of short-term events and long-term potential causes of the event, it can control the number of feature extraction, reducing the amount of calculation, to avoid the emergence of over-fitting phenomenon, thus increasing the accuracy of the analysis of the causal effect.