Electric power material demand prediction method based on spatial-temporal clustering

The invention discloses an electric power material demand prediction method based on spatial-temporal clustering, and the method comprises the steps: calling and cleaning historical demand data of electric power materials; extracting new features of cleaned data samples in the time dimension; carrying out replacement processing on noise in the cleaned historical demand data of the electric power materials, and smoothing data noise; clustering the data subjected to replacement processing in time and space dimensions, and dividing the clustered data into a training set and a test set; obtaining reference materials adjacent to new materials according to the new features by using a k-nearest neighbor algorithm, and updating intra-class points; training the training set by using a plurality of training models, and predicting by using test set data; measuring the prediction precision of the training models in the target class by using model precision evaluation indexes, and selecting the training model with the hig