Research On Data Remediation Method Of Marketplace Passenger Data Missing

In the fields of marketplace, passenger information is an important factor for its operating. Monitoring, researching and analyzing passenger information, then making decision, positioning, using various means to attract passengers, ultimately enhance the capacity of competition.But, inevitably,monitoring will cause data missing problem by some malfunction.In the problem this paper studied,using inductive laser technology for passenger monitoring may possibly cause a mass of consecutive data missing,which is called"large window"passenger data missing.The goal of this paper is to give a feasible solution for this issue.First,this paper introduced the conception of missing data and several traditional data missing remediation methods.Then for the problem studied in the passenger data monitoring and analyzing system,a forecasting remediation model is constructed.After simple introducing of BP neural network algorithm,the paper construct a remediation model based on BP neural network, and use it to"large window"passenger data missing problem.It is indicated that though the result could simulate the trend of data,it has large error.Considering that passenger has moving quality, the training data is pre-adjusted based on time,after data relativity analysis,the result of BP neural network is improved and the precision is promoted.It is indicated that the pre-adjustment is effective.Because the precision and stability of single forecasting model could not be ensured, and that combination forecasting model could use different single model information,so a combination forecasting model based on multianalysis of linear regression forecasting model and BP neural network forecasting model is constructed,its weights is gived by another non-linear BP neural network.Finally,for promoting the model's precision,error is also forecasted,the result is corrected.The experiments indicate that combination forecasting model based on error-correcting could promote the precision again,and excel any single model,so it is feasible to use this model to"large window"passenger data missing problem.