Incomplete Data Classification Algorithm And Its Application In Building Energy Consumption Prediction

For incomplete data,has developed many preprocessing methods,such as deletion and filling.Deletion will cause information waste and filling method will bring uncertainty,so incomplete data classification is called the research focus.The prediction of building energy consumption is of great significance to the construction of an environment-friendly society.In the prediction of building energy consumption,all kinds of data loss make the traditional method difficult to apply.Incomplete data classification and its application in the prediction of building energy consumption have important research significance.In this paper,incomplete data classification and its application in building energy consumption prediction are studied.In view of the problem of incomplete data classification,two algorithms of incomplete data classification are proposed,and the effectiveness of the algorithm is verified.On the basis of the algorithm,it is applied in building energy consumption prediction to verify the feasibility of the application of the algorithm.The specific research contents are as follows:(1)An incomplete data classification algorithm based on D-S evidence theory is proposed.Combining clustering and classification,clustering and selecting training sets for complete objects,building multiple training sets and training corresponding classifiers according to the existing attributes of the remaining objects;Under the D-S evidence theory,some objects are divided into meta-classes,and the incomplete objects in metaclasses are estimated,classified and fused to get the final classification results.The algorithm can effectively represent imprecise objects and reduce the error rate.(2)An incomplete data classification algorithm based on attributes is proposed.The incomplete data set is divided into several sub data sets,each sub data set trains a classifier and calculates the classifier weight;The attribute similarity and attribute weight are integrated to estimate the missing value of the test set,and the classifier is used to classify the test set and fuse the classification results.This algorithm can effectively reduce the misclassification rate and has a high applicability to the selection of the basic classifier.(3)Based on the proposed incomplete data classification algorithm,two corresponding building energy prediction model is constructed to predict the building energy consumption.By comparing with the three indicators of the other two methods,the incomplete data classification algorithm proposed in this paper can be effectively used in building energy consumption prediction.