Method and system for improving feature generation efficiency of machine learning sample

The invention provides a method and a system for improving feature generation efficiency of a machine learning sample. The method comprises the steps of obtaining a flow table in which data records with time sequences are recorded, one row of the flow table corresponding to one data record, and one column of the flow table corresponding to one field; and for the timing sequence characteristics needing to be counted, carrying out first coarse-grained first-order aggregation on corresponding fields in the pipeline table, storing a first-order aggregation result, and then obtaining the timing sequence characteristics of the machine learning sample corresponding to each data record in the pipeline table based on the stored first-order aggregation result. According to the method and the system, the generation efficiency of the time sequence characteristics can be improved. 提供了一种提高机器学习样本的特征生成效率的方法及系统。所述方法包括：获取记录有具有时序的数据记录的流水表，其中，所述流水表的一行对应一条数据记录，所述流水表的一列对应一个字段；针对需要统计的时序特征，先对所述流水表中的相应字段进行粗粒度的一阶聚合并保存一阶聚合结果，再基于所保存的一阶聚