Data mining model for an oriental medicine via BYY Binary Independent Factor Analysis was proposed. This model consists of a three layered structure that has two type learning algorithm in two steps. In the first step BYY harmony learning and model selection are provided for Binary Independent Factor Analysis on networks with one hidden layer binary units. In the second step, the supervised learning algorithm is provided. We develop an automatic oriental medical diagnostic system based on BYY Binary Independent Factor Analysis. Moreover, experiments are made on the application to the diagnostic problem of 'liver disease' with the medical sampling data in the oriental medical diagnostic perspectives.