Property prediction system

A method for predicting a property of a light-emitting device, the method comprising:

• inputting first data and second data,
• generating a learning data set comprising first learning data and second learning data, the first learning data generated from the first data and the second learning data generated from the second data,
• performing supervised learning on the basis of the learning data set,
• inputting third data,
• generating data used for predicting the property of the light-emitting device from the third data,
• making an inference of at least one of initial properties and reliability of the light-emitting device from the data used for predicting the property of the light-emitting device on the basis of the learning result of the supervised learning, and
• outputting fourth data including at least one of the initial properties and the reliability of the light-emitting device,
• wherein the first learning data includes information of a structure of a first light-emitting device, information of a first organic compound included in the first light-emitting device, information of initial properties of the first light-emitting device, and information of reliability of the first light-emitting device,
• wherein the second learning data includes information of a structure of a second light-emitting device, information of a second organic compound included in the second light-emitting device, information of initial properties of the second light-emitting device, and information of reliability of the second light-emitting device, and
• wherein the data used for predicting the property of the light-emitting device includes information of a structure of the light-emitting device and information of an organic compound included in the light-emitting device.