Probabilistic Airport Traffic Demand Prediction Incorporating the Weather Factors

Abstract

With the development of air transport industry in China, the congestion problem in the terminal areas of busy airports has become increasingly serious. In order to alleviate the increasingly frequent air traffic congestion, it is necessary to accurately and objectively predict traffic flow. Traditionally, most predicted methods are based on the number of aircrafts flight in the terminal area to obtain deterministic traffic flow data, without considering the impact of uncertain factors on the prediction results. Based on the uncertainty of demand, this paper uses a probability density prediction method based on quantile regression neural network and kernel density estimation, to analyse the variation of traffic flow at different quantiles according to the obtained continuous conditional quantile function. Predicting the probability density of traffic flow on a certain day, and then comparing the point prediction value corresponding to the peak value, which consider the weather factor and the conditional probability density prediction curve without considering the weather factor, it is concluded that considering the weather factor can make the traffic flow prediction more accurate.

Modeling Censored Mobility Demand Through Censored Quantile Regression Neural Networks

“…In studies of the electric power industry, He et al use non-censored QNN to estimate latent distributions of electricity production [37], [38] and consumption [39], while [40] and [41] use non-censored QNN to predict electricity loads. In the transport domain, [42] use a non-censored QNN with a single hidden neuron to predict 15min air traffic in a Chinese airport, and [43] devise a non-censored, multi-output QNN that jointly estimates mean and quantiles, whereby they predict taxi demand in New York City, in 30 minutes intervals.…”

Machine learning methods for transportation under uncertainty

“…In the Transport domain,(Tian et al, 2020) use a non-censored QRNN with a single hidden neuron to predict 15 min air traffic in a Chinese airport, and Rodrigues and Pereira (2020) devise a non-censored, multi-output QRNN that jointly estimates mean and quantiles, whereby they predict 30 min taxi demand in New York City.Very few works apply QRNN in a Censored setting (CQRNN) Cannon (2011). develops a general architecture for both censored and non-censored QRNN, which he implements as an R package.…”