To provide an optimum hyper parameter for determining a learning model using a natural language as a target. An optimization apparatus including: a processor and a memory and performing learning of a document set by natural language processing has an optimization section configured to determine a hyper parameter satisfying a predetermined condition on the basis of previously set group data, generate a learning model by the determined hyper parameter, and acquire a high-dimensional vector from the learning model; and a high-dimensional visualization section configured to convert the high-dimensional vector of a word or document as an analysis target on the basis of the group data.