The invention discloses a data set determination method and device. The method includes the steps: receiving commands for indicating to target data sets from a plurality of first data sets; determining target probabilities of instance data in each first data set according to a first attribute set; determining acquisition quality information of each first data set based on the target probabilities of all instance data in each first data set; determining the target data sets for data analysis. Data of the target data sets are used for data analysis, the target probabilities are probabilities of the instance data belonging to target types, the first attribute set comprises data attributes for indicating the target types, the acquisition quality information is used for indicating the qualities of the first data sets acquired according to the target types, and the acquisition quality information of the target data sets meets preset quality requirements in the first data sets. The data set determination method solves the technical problem that data sets with high quality cannot be acquired in related technology.