Material demand prediction method based on feature extraction and improved random forest

The invention relates to a material demand prediction method based on feature extraction and an improved random forest, and the method comprises the steps: importing material demand data in historical project data and project attribute data corresponding to material demands, and carrying out the data extraction, and obtaining a project material historical data set; based on the established industry and engineering classification knowledge base, grouping the project material historical data set; sorting the importance of the materials in the historical projects according to the project attribute data, and screening out the types of the materials needing to be predicted; training a corresponding random forest model by using the material grouping historical data corresponding to each material type; and inputting the types of materials needing to be predicted, obtaining project attribute data of the materials, and importing the corresponding random forest model to predict the demand quantity of the materials. The