A Fuzzy Support Vector Machine with Weighted Margin for Flight Delay Early Warning

Flight delay early warning can reduce the negative impact of the delay. Determining the delay grade of each interval is essentially a multi-class classification problem. This paper presents a flight delay early warning model based on a fuzzy support vector machine with weighted margin (WMSVM) , which adjust the penalties to samples and the margins between samples and the hyperplane according to the fuzzy membership to produce a more reasonable optimal hyperplane. Through one-against-one (OAO) method, the original FSVM is extended to solve multi-class classification problem .Experiments show that the method used to establish the early warning model can predict the delay grade effectively and also prove that the OAO-WMSVM has better performance than OAO-SVM.