A data processing apparatus includes a connecting unit that distributes the plurality of processing modules over the stages, and connects the plurality of processing modules such that a plurality of partial data are processed in parallel. The data processing apparatus detects, with respect to at least a part of the stages, a ratio of an amount of data for which processing in the subsequent stage has been executed, as a passage rate, acquires a processing time for a data amount to be processed in each stage, for which the passage rate was detected, based on the passage rate, and determines the number of processing modules distributed to each stage based on the data amount.