Predictive modeling of inbound demand at major European airports with Poisson and Pre-Scheduled Random Arrivals

Abstract

This paper presents an exhaustive study of the arrivals process at eight major European airports. Using inbound traffic data, we define, compare, and contrast a data-driven in-homogeneous Poisson and Pre-Scheduled Random Arrivals (PSRA) point process with respect to their ability to predict future demand. As part of this analysis, we show the weaknesses and difficulties of using a non-homogeneous Poisson process to model the arrivals stream. On the other hand, our novel and simple data-driven PSRA model captures and predicts the main properties of the typical arrivals stream with good accuracy. These results have important implication for the modeling and simulation-based analyses of inbound traffic and can improve the use of available capacity, thus reducing air traffic delays. In a nutshell, the results lead to the conclusion that, in the European context, the PSRA model provides more accurate predictions.

Smart Public Transport: A Bi-Objective Model for Maximizing Synchronizations and Minimizing Costs in Bus Timetables

“…Concerning the proposed bus occupancy model, it combines random passenger boarding with deterministic alighting processes. The random boarding process is based on a Poisson distribution, which is a probabilistic distribution that is a commonly used model for random arrivals that are independent and identically distributed [27,28]. Given the assumption of regularity throughout the planning period-during which relevant data for the problem can be considered steady-the Poisson process is considered to have a fixed rate.…”

Variations in urban traffic

“…We refer to Goh and Barabási (2008), Kim and Whitt (2014), Lancia and Lulli (2020), and Paxson and Floyd (1995) for similar observations in different contexts. In the following subsection, we discuss earlier approaches to capture the above-mentioned characteristics.…”

“…Based on literature (Chen et al, 2012;Guo & Williams, 2012;Thomas et al, 2008), we assume that the noise variance 2 (s d,t ) scales linearly with volume s d,t , i.e., the (conditional) noise variance is a linear function of the underlying systematic volume. Although different mechanisms exist to estimate s from x (Habtemichael & Cetin, 2016;Lancia & Lulli, 2020;Thomas et al, 2008), in the next section we use an approach that explicitly accounts for volume-dependent noise as further elaborated upon in Chapter 4.…”