In order to further improve the accuracy of the variety determination and consumption quantity prediction of carried aircraft spare parts, the problem of feature selection in the nonlinear multi-influence factors variety classification and consumption prediction model was studied. The factors affecting the demand for spare parts of carrying aircraft are analyzed, a 3-level feature system is established, and corresponding feature sets are extracted according to the variety determination and consumption prediction. XGboost, grey relation analysis (GRA), decision making trial and evaluation laboratory (DEMATEL) were used to rank the importance degree and analyze the correlation of each influence feature. Synthetically use qualitative and quantitative analysis methods to select features. The feature sets of the simplified edition which can be used for species determination and quantity prediction are established respectively. This study can provide reference for improving the accuracy and computing efficiency of the identification and prediction of spare parts for carrying aircraft.