Alarming Large Scale of Flight Delays: an Application of Machine Learning
AbstractThis chapter shows an example to use machine learning method in applications. Limited by the space of pages, some details have been skipped to get the main point. As a supplement, we will show some reference in which the details could be found. The flight delays will cause a series of serious consequences. However, it is very difficult to predict flight delays accurately. Some models have been presented to describe the flight delays, such as Milan Janic's disruption model (Janic, 2005), Ning Xu's Bayesian network model (Xu, 2007) and Zonglei Lu's decision tree model (Lu et al, 2008a) and so on. However, there is no model could predict the flight delays accurately up to now. These models could give only some reference of the prediction. There are some machine learning methods mentioned in this chapter. The systematic introduction about these methods could be found in the classical machine learning books, such as (Mitchell, 1997) and (Witten & Frank, 2005). For more details, the k-means clustering algorithm is first mentioned in (Jain, 1967). The Dunn's index, Davies-Bouldin's index, CS index and Lu's index has been introduced by (Dunn, 1974), (Davies & Bouldin, 1979), (Chou et al, 2004) and (Lu et al, 2008b), respectively. Some details about these clustering validity indexes could also be found in (Halkidi et al, 2002), a survey of clustering validity index. The C4.5 decision tree model, the naive Bayes model and the BP neural network model is mentioned in (Quinlan, 1993), (John & Langley, 1995), (Werbos, 1974), respectively.