Research On Data Of PAP Based On Machine Learning

With the steady progress of reforming and strengthening the army,the informationization of the army is accelerating,and the data volume of the major war zones,arms and Military Commission organs is growing rapidly.As one of the army data,armed police data has a broad application prospect,but because of confidentiality restrictions,high sensitivity and difficult access to the Internet,it has not been paid attention to and rationally utilized at all levels.Based on in-depth understanding of domestic and foreign police data analysis and research,this paper reasonably uses LSTM neural network.The network model,which trains the processed data of the armed police,is used to predict the number of major events that endanger social stability in a certain area in the future.Specific research contents are as follows:1.The traditional data storage methods of armed police affairs are extensive and diverse,complex and cumbersome.Data sources involve many fields,have many categories and different structures,and have a lot of useless information.In this paper,the police data is sorted and classified,and the missing values in the data are processed by the deletion method and the filling method,one-hot coding is used to deal with attribute classification for discrete data,and data generalization method is used to improve data consistency.By cleaning and filtering theinformation in existing data,data storage is standardized and normalized.The level at which the prediction system can be applied.2.This paper collects and processes the original data of the Armed Police and applies it to the training of LSTM neural network model.In this paper,24 different parameters are set up,and different tests are carried out.Finally,a relatively valuable prediction model is obtained.