To improve accuracy of determining a state pertaining to a determination device, a determination program and a determination method.SOLUTION: The device generates a sparse structure model from time-series learning data obtained by a plurality of sensors in a learning stage, and determines a state of target data on the basis of time-series target data obtained from the plurality of sensors in an operation state and the sparse structure model. In a learning stage (1), the learning data is clustered into a plurality of aggregates that are equivalent to a plurality of normal modes (3). Meanwhile, the sparse structure model in which a relationship between parameters in the learning data included in each of the plurality of aggregates is represented by a Markov random field is created for each aggregate (4). Also, a target aggregate, among the plurality of aggregates, which is the target of comparison with the target data is an operation stage (7) is estimated (8). Then, the state of the target data is determined o
