The present invention is provided with: a processing data acquisition unit (11) that acquires a plurality of items of processing data including data related to a time-series processing condition for causing a processing machine (4) to perform processing for a plurality of steps, and data related to the state of the processing performed for the plurality of steps by the processing machine (4) on the basis of the processing condition; an analysis unit (13) that calculates levels of importance for the individual items of processing data; an extraction unit (14) that extracts, from among the plurality of items of processing data on the basis of the levels of importance calculated by the analysis unit (13), a plurality of items of transfer learning data for performing transfer learning of a first machine learning model; and a learning unit (153) that uses the plurality of items of transfer learning data to perform the transfer learning of the first machine learning model and generates a second machine learning model that uses, as an input, data related to the time-series processing condition for the plurality of steps to output data related to the state of the processing after the plurality of steps have been performed.