Representation learning method of spatial position information based on convolutional auto-encoder

The invention discloses a spatial position information representation learning method based on a convolutional auto-encoder, and the method comprises the following steps: S1, data acquisition and data preprocessing: to-be-studied data is acquired, the data is table-type data and needs to contain a data unique identifier, timestamp information, latitude and longitude information and N pieces of additional service feature information, and N is a positive integer; and carrying out normalization processing on the obtained N service features. According to the method, on the basis of the geographic position information, the neighbor relation on the geographic position can be better represented by introducing a related spatial data processing method and the convolutional auto-encoder, and the method can be better integrated with a modeling method of machine learning through the processing result of the encoder, so that the original rule derivation method and thought are expanded, and the method is more efficient and