Research On NOSHOW Rule Discovery Based On Civil Aviation Passenger Service Data

In the aviation market,the phenomenon that passengers cannot be scheduled after reservations(NOSHOW)will eventually lead to wasted seats,which causes huge economic losses to the airlines.With the rapid development of big data technology,how to extract effective rules from CKI(Check-In,departure data)generated in the massive civil aviation passenger service systems to reduce seat and revenue loss has become a major problem for major airline.Firstly,this paper analyzes the CKI data generated by the civil aviation passenger service system,and concludes that this data set is a massive imbalanced data set.For this data feature,the C5.0 data mining algorithm is applied to process large data sets,and the initial NOSHOW classification model is constructed to verify the feasibility of the decision tree algorithm in the civil aviation passenger NOSHOW classification.The problem of low classification speed and high memory occupancy performance is solved.Secondly,according to the experimental verification constructed by the initial classification model,it is found that the impact of the different classification costs in practical applications is different.To address this type of situation,a NOSHOW classification rule discovery method based on Misjudgment costs is proposed.The method takes account of the misjudgment cost of different cases,and effectively reduces the impact of high cost misjudgment on the basis of guaranteeing that the total classification error does not change much.Thirdly,based on the classification model constructed above,a strong factor that has a large impact on NOSHOW is obtained.Due to the excavated NOSHOW classification rules,the correlation between strong factors was not considered.Therefore,using the optimized Apriori method to extract the strong factor association rules for the above-mentioned extracted NOSHOW strong factors,the implicit relation between the NOSHOW strong factors was revealed.Finally,by conducting experiments and research on massive CKI data,this research method constructs a low-cost,high-efficiency NOSHOW classification model and produces an intuitive and easy-to-understand rule set,which provides effective decision-making basis for each major airline to achieve accurate NOSHOW classification and revenue improvement management.