A method for predicting grain purchases based on farmer portraits

Abstract

The present invention discloses a method for predicting grain purchases through farmer portraits, and relates to the technical field of data labels. The method comprises converting collected original farmer basic information into structured data to obtain original data labels, extracting fact labels according to fact data attributes based on the original data labels, constructing model prediction labels based on the fact labels, constructing a farmer portrait label system through the original data labels, the fact labels and the model prediction labels, setting the effective time limit of the labels in the farmer portrait label system according to the differences between different labels, the label change frequency and the label stability, training a grain purchase prediction model based on the set labels using the farmer behavior data corresponding to the labels, and using the grain purchase prediction model to predict grain purchases.

Images