Airline Delays Research Articles

Airline delay prediction by machine learning algorithms

Flight planning, as one of the challenging issue in the industrial world, is faced with many uncertain conditions. One such condition is delay occurrence, which stems from various factors and imposes considerable costs on airlines, operators, and travelers. With these considerations in mind, we implemented flight delay prediction through proposed approaches that are based on machine learning algorithms. Parameters that enable the effective estimation of delay are identified, after which Bayesian modeling, decision tree, cluster classification, random forest, and hybrid method are applied to estimate the occurrences and magnitude of delay in a network. These methods were tested on a U.S. flight dataset and then refined for a large Iranian airline network. Results showed that the parameters affecting delay in US networks are visibility, wind, and departure time, whereas those affecting delay in Iranian airline flights are fleet age and aircraft type. The proposed approaches exhibited an accuracy of more than 70% in calculating delay occurance and magnitude in both the whole-network US and Iranian. It is hoped that the techniques put forward in this work will enable airline companies to accurately predict delays, improve flight planning, and prevent delay propagation.

Airport delays and metropolitan employment

AbstractThis study provides the first empirical assessment of the impact airline delays have on urban employment. While previous works have suggested that road congestion can slow down regional development, the influence of air traffic delays on metro‐level jobs has not been examined comprehensively. The present study uses a nine‐year panel of quarterly data, which covers passenger airline traffic and delays at airports across urban areas in the United States. The panel also includes data on total and industry‐specific employment at the metropolitan‐area level. Our empirical estimates of the impact of air traffic on total employment are comparable to previously reported measures in the literature. However, we find that service‐sector employment is less sensitive to air traffic than other studies suggested. We provide new evidence confirming that delays have a negative effect on employment, a finding that is robust to various specifications of our empirical model. Our results indicate that a 10 percent increase in the number of delayed flights leads to up to a 0.15 percent decrease in total and service‐sector employment, a 0.47 percent decline in leisure and hospitality employment, and a 0.7 percent reduction in the employment level of goods‐producing jobs.

Online updating method with new variables for big data streams.

For big data arriving in streams, online updating is an important statistical method that breaks the storage barrier and the computational barrier under certain circumstances. In the regression context, online updating algorithms assume that the set of predictor variables does not change, and consequently cannot incorporate new variables that may become available midway through the data stream. A naive approach would be to discard all previous information and start updating with new variables from scratch. We propose a method that utilizes the information from earlier data in the online updating algorithm with bias corrections to improve efficiency. The method is developed for linear models first, and then extended to estimating equations for generalized linear models. Closed-form expressions for the efficiency gain over the naive approach are derived in a particular linear model setting. We compare the performance of our proposed bias-correcting approach and the naive approach in simulation studies with data generated from a normal linear model and a logistic regression model. The method is applied to a study on airline delay, where reasons for delays were only available more recently, starting in 2003.

Big Data Analytics on Aviation data for the prediction of Airline Trends in Seasonal Delay

The trajectory of the global airline industry is pretty much like that of an aircraft. At times it takes off for the high skies and at times, it dips to ground levels. In between these highs and lows, lies the story of the industry – of its survival, of the new and emerging trends that fuel its growth. The issue of airline delays, as measured by the number of late arrivals as a percent of total operations, has been of increasing importance in recent years as most of the population chooses air travel as a preferred mode of transportation. This study provides the result about the total flight delay for a specific period of time caused due to climate, security, carrier, NAS, Arrival and Departure based on total number of flights getting delayed over the past few years (2006, 2007 and 2008). The historic data which is to be analysed is stored on the databases such as MongoDB and Hive. The usage of time series analysis along with the integration of heterogeneous database helps to achieve the Airline Seasonal Delay which is implemented and visualized in R. The reports are generated by using time series modeling to provide the insights for the aviation industry to take future measures to avoid delays and manage them.

Airline delay management problem with airport capacity constraints and priority decisions

This paper deals with the Airline Delay Management Problem (ADMP), which can be described as the task of dealing with daily airline operational delays and deciding whether to delay subsequent flights at a hub airport or to have them departing on time. An innovative integer linear programming approach is presented to the capacitated case of the ADMP and airport limitations in terms of bay availability, taxiway capacity and runway separation are incorporated to represent capacity constraints. Fuel cost, passenger compensation, and passenger inconvenience costs are included in the objective function. The decision variables include the re-timing of flight departures and arrivals, the use of the airport capacity over time and the rebooking of passengers in case of missed connections. To guarantee the linearity of the optimization model and fast computational times, a receding horizon modeling framework is adopted. The approach is applied to a case study using real operational and passenger data from an international hub-and-spoke carrier. The case study shows the capability of the linear model to deal with a complete day of operations within a few minutes. The results suggest that the proposed approach can lead to cost reductions of almost 30% during recovery, when compared with the solution from the airline. In addition, a sensitivity analysis is provided to investigate the impact of not including passenger inconvenience costs and of reducing runway capacity.

Predicting Fuel Consumption and Flight Delays for Low-Cost Airlines

Low-cost airlines (LCAs) represent a new category of airlines that provides low-fare flights. The rise and growth of LCAs has intensified the price competition among airlines, and LCAs require continuous efforts to reduce their operating costs to lower flight prices; however, LCA passengers still demand high-quality services. A common measure of airline service quality is on-time departure performance. Because LCAs apply efficient aircraft utilization and the time between flights is likely to be small, additional effort is required to avoid flight delays and improve their service quality. In this paper, we apply state-of-the-art predictive modeling approaches to real airline datasets and investigate the feasibility of machine learning methods for cost reduction and service quality improvement in LCAs. We address two prediction problems: fuel consumption prediction and flight delay prediction. We train predictive models using flight and passenger information, and our experiment results show that our regression model predicts the amount of fuel consumption more accurately than flight dispatchers, and our binary classifier achieves an area under the ROC curve (AUC) of 0.75 for predicting a delay of a specific flight route.

Integrated Disruption Management and Flight Planning to Trade Off Delays and Fuel Burn

In this paper we present a novel approach addressing airline delays and recovery. Airline schedule recovery involves making decisions during operations to minimize additional operating costs while getting back on schedule as quickly as possible. The mechanisms used include aircraft swaps, flight cancellations, crew swaps, reserve crews, and passenger rebookings. In this context, we introduce another mechanism, namely flight planning that enables flight speed changes. Flight planning is the process of determining flight plan(s) specifying the route of a flight, its speed, and its associated fuel burn. Our key idea in integrating flight planning and disruption management is to adjust the speeds of flights during operations, trading off flying time (and along with it, block time) and fuel burn; in combination with existing mechanisms, such as flight holds. Our goal is striking the right balance of fuel costs and passenger-related delay costs incurred by the airline. We present both exact and approximate models for integrated aircraft and passenger recovery with flight planning. From computational experiments on data provided by a European airline, we estimate that the ability of our approach to control (decrease or increase) flying time by trading off with fuel burn, as well as to hold downstream flights, results in reductions in passenger disruptions by approximately 66%–83%, accompanied by small increases in fuel burn of 0.152%–0.155% and a total cost savings of approximately 5.7%–5.9% for the airline, may be achieved compared to baseline approaches typically used in practice. We discuss the relative benefits of two mechanisms studied—specifically, flight speed changes and intentionally holding flight departures, and show significant synergies in applying these mechanisms. The results, compared with recovery without integrated flight planning, are because of increased swap possibilities during recovery, decreased numbers of flight cancellations, and fewer disruptions to passengers.

Socioeconomic effect of dust storms in Kuwait

Kuwait is an arid country with an annual average rainfall of about 110 to 120 mm. Hence, dust storms have become a common environmental crisis. Although the dry summer months commonly have more dust storms, in recent years, dust storms occur more frequently in the spring season as well. Accumulated data of dust storms in Kuwait for the past 14 years (2001–2014) showed that the month of March had the highest number of dust storms (total 19), which is rather unusual, with an average of 8 dust storms per year (year 2008 had the highest dust storms of total 22). This study explored four socioeconomic effects of dust storms in Kuwait, specifically traffic accident rates, oil export loss due to close out of marine terminals, airline delays due to airport operation shutdown, and agriculture degradation. Statistical analysis using t test and Pearson correlation showed no apparent relationship between dust storms and traffic accident rates or agricultural production; however, loss of oil export and flight delay cost were affected by dust storms. There has been very few published research on the socioeconomic impact of dust storms; this is the first paper that explores the detailed socioeconomic effect of dust storms in Kuwait.

Airline delays, congestion internalization and non-price spillover effects of low cost carrier entry

This paper develops an econometric model of flight delays to investigate the influence of competition and dominance on the incentives of carriers to maintain on-time performance. We consider both the route and the airport levels to inspect the local and global effects of competition, with a unifying framework to test the hypotheses of 1. airport congestion internalization and 2. the market competition-quality relationship in a single econometric model. In particular, we examine the impacts of the entry of low cost carriers (LCC) on the flight delays of incumbent full service carriers in the Brazilian airline industry. The main results indicate a highly significant effect of airport congestion self-internalization in parallel with route-level quality competition. Additionally, the potential competition caused by LCC presence provokes a global effect that suggests the existence of non-price spillovers of the LCC entry to non-entered routes.

Airline Delays, Congestion Internalization and Non-Price Spillover Effects of Low Cost Carrier Entry

This paper develops an econometric model of flight delays to investigate the influence of competition and dominance on the incentives of carriers to maintain on-time performance. We consider both the route and the airport levels to inspect the local and global effects of competition, with a unifying framework to test the hypotheses of 1. airport congestion internalization and 2. the market competition-quality relationship in a single econometric model. In particular, we examine the impacts of the entry of low cost carriers (LCC) on the flight delays of incumbent full service carriers in the Brazilian airline industry. The main results indicate a highly significant effect of airport congestion self-internalization in parallel with route-level quality competition. Additionally, the potential competition caused by LCC presence provokes a global effect that suggests the existence of non-price spillovers of the LCC entry to non-entered routes.

Losses Loom Longer than Gains: Modeling the Impact of Service Crises on Perceived Service Quality over Time

Service providers sometimes face mass service failures. These problems occur across service industries, ranging from severe Internet outages to major delays for airlines or trains. The literature has not yet addressed the following key question: How do service crises affect perceived service quality (PSQ) over time? To answer this question, the authors introduce a Double-Asymmetric Structural Vector Autoregressive model. It captures not only the short- and long-term effects of objective service performance on PSQ but also the differential effects of service crises versus service restoration. The authors analyze a unique data set from a major European railway company, spanning seven years of monthly observations. During this period, severe winter weather caused dramatic service crises. The authors find that performance losses loom larger than gains in the short run and also have permanent negative effects on PSQ in the long run. Consequently, a crisis followed by a restoration will result in a net negative long-term effect on PSQ. The impact of a crisis also depends on the prior trend in objective service performance.

Low-cost carrier competition and airline service quality in Europe

The authors are investigating whether a higher presence and more efficient operations of low-cost carriers (LCCs) can increase the service quality in terms of on-time performance of all the flights landing at an airport. We sample 100 European airports located in 76 metropolitan areas of diverse sizes in 19 countries on both a daily and flight-by-flight basis during the period from April 2011 to December 2012. We construct a panel dataset at the flight code level comprising about 3.5 million observations. We find that LCCs contribute to a reduction of delays for airlines and flights landing at the observed airport. From the customers’ point of view and taking into consideration the level of service, we conclude that the presence of LCCs represents a positive externality for an airport. Airport management may therefore consider the proactive increase of LCCs market share in their long-term business strategies.

A two-stage stochastic integer programming model for air traffic flow management

The high cost of flight delays for airlines has motivated scientific research in air traffic flow management (ATFM). The majority of ATFM models in the literature are deterministic and do not take into account stochastic factors such as weather. In this paper, new stochastic programming models for ATFM are proposed. The models include as tactical control options: ground holding, airborne holding and rerouting. To solve the models, a new heuristic method that takes advantage of the problem structure is derived and illustrated. Computational results show that the heuristic method provides practical computation times. Furthermore, the value of the stochastic solution is up to 14% for cases where adverse weather affects a significant part of the network. This implies that using the proposed approach to make air traffic flow decisions can lead to tangible monetary benefits for airlines.

Is Your Flight Really on Time?

The majority of studies on airline delays have been based on databases that contain information only about the final delay outcome associated with a flight; that is, did a flight arrive early or late at the gate and, if so, by how many minutes? This paper uses a database from FlightStats that contains information on the timing of airline notifications of delays to customers. Specifically, the delay notification strategies of six U.S. airlines are compared, and the timing and accuracy of their delay forecasts are analyzed with data from 2010. Results show that airlines use different delay notification strategies. AirTran and Southwest announce delays more frequently throughout the flight departure process, whereas American Airlines, American Eagle, Delta, and JetBlue announce the majority of their delays in two distinct periods before departure. Results also indicate that “no news is bad news” in the sense that the closer to departure that a delay is first announced, the more likely it is to be a long delay. An assessment of forecasting accuracy reveals that the majority of airlines consistently underforecast delays; the two notable exceptions are JetBlue and AirTran. The timing and accuracy of delay forecasts has important implications for airports, particularly as it relates to their ability to develop contingency plans for irregular operations.

Airline Delay Time Series Differentials

Flight delays affect passenger travel satisfaction and increase airline costs. The authors explore airline differences with a focus on their delays based on autoregressive integrated moving averages. Aviation daily data were used in the analysis and model development. Time series modelling for six airlines was done to predict delays as a function of airport's timeliness performance. Findings show differences in the time series prediction models by airline. Differential analysis in the time series prediction models for airline delay suggests variations in airline efficiencies though at the same airport. The differences could be attributed to different management styles in the countries where the airlines originate. Thus, to improve airport timeliness performance, the study recommends airline disaggregated studies to explore the dynamics attributable to determinants of airline unique characteristics.

Valuating Surface Surveillance Technology for Collaborative Multiple-Spot Control of Airport Departure Operations

Airport departure operations are a source of airline delays and passenger frustration. Excessive surface traffic is a cause of increased controller and pilot workload. It is also a source of increased emissions and delays, and does not yield improved runway throughput. Leveraging the extensive past research on airport departure management, this paper explores the environmental and safety benefits that improved surveillance technologies can bring in the context of gate- or spot-release strategies. The paper shows that improved surveillance technologies can yield 4% to 6% reduction of aircraft on taxiway, and therefore emissions, in addition to the savings currently observed by implementing threshold starategies under evaluation at Boston Logan Airport and other busy airports during congested periods. These calculated benefits contrast sharply with our previous work, which relied on simplified airport ramp areas with a single departure spot, and where fewer environmental and economic benefits of advanced surface surveillance systems could be established. Our work is illustrated by its application to New-York LaGuardia and Seattle Tacoma airports.

Competition and service quality: New evidence from the airline industry

Previous studies examine the relationship between competition and airline service quality by regressing on-time performance on market structure. These studies implicitly assume that market structure is exogenously determined with respect to service quality. To address the likely endogeneity of market structure I employ two distinct instrumental variables. The first is lagged market structure. The second exploits a global airline merger that induced differential variations in market structure across hundreds of airport-pair markets. I find that the effect of competition on airline delays is three times stronger than previous studies suggest.

Using Ant Algorithm to Arrange Taxiway Sequencing in Airport

Over the last decade, air traffic flow at airports experienced an increase never seen before. Whereas several improvements have been achieved in enlarging the en-route traffic capacity, little has been done in order to decrease congestion on the airport surface. Aircraft taxiing is an expensive process, responsible for causing delays for both passengers and airlines and pollution problems. This paper presents an optimization solution for aircraft taxi-scheduling problem using Ant Colony Optimization, a method for search and optimization inspired by the behavior of real ants in nature. The simulations presented were based on real data from Brasilia International Airport. However, the developed model can be used in any airport for optimal taxiing routes searching. The paper has shown the optimization of taxiing to be efficient in achieving minimization of aircraft taxi time and complies with security constraints at airports. Index terms – Ant algorithm, air traffic flow management, optimization, simulation, taxiway sequencing

Special issue: Selected papers from the Innovative Research Workshop 2010

Over nearly 10 years the Innovative Research Workshop and Exhibition held annually in early December at the EUROCONTROL Experimental Centre in Bretigny-sur-Orge, France, has brought together researchers and practitioners from air traffic management and related areas. The workshop focused specifically on innovative research and has continuously attracted inspiring new ideas, studies and concepts. The 2010 edition of the INO Workshop was a notable event (not least because snow kept the attendants locked into the centre one evening) and we have decided to dedicate a special edition of the Journal of Aerospace Operations to it. We have selected what we believe are the most interesting and innovative amongst the papers submitted for presentation at the conference and invited the authors to prepare an extended version of their original manuscript. There are seven articles in this edition which we are sure the reader will find stimulating and rewarding. “A quantitative exploration of flight prioritisation principles, using new delay costs” by Andrew Cook and Graham Tanner proposes a systematic approach to assessing the cost of delays for airlines and how these findings should be applied to flight prioritisation procedures instead of the simple application of delay minutes for optimisation purposes. “Analysis of the ecological and economic impact of a Single European Sky by simulating Freeflight trajectories with Lido/Flight” by Andreas Henn, Urban Weishaar and Philipp Bock proposes a method for quantifying the fuel saving potential of user preferred trajectories as compared to the present trajectories in the European airspace. The application of the flight planning software presently used by Lufthansa and several other airlines to a historic data set shows a fuel saving potential of more than 3% which translates to economic and ecological benefits alike. “4D trajectory management using Contract of Objectives” by Sandrine Guibert and Laurent Guichard describes a series of experiments carried out in order to assess the benefits and usability of the

Estimating domestic US airline cost of delay based on European model

Researchers are applying more holistic approaches to the feedback control of the air transportation system. Many of these approaches rely on economic feedback, including the cost of delays to the airlines. Establishing an accurate mechanism for estimating the cost of a delays for each portion of a flight (gate costs, taxiing in and out costs, and en-route costs) is useful for many aspects of modeling airline behavior and for better understanding the likely impact of regulations.Cook et al. (2004) developed a rigorous methodology and collected data for estimating the components of airline delay costs for various segments of a scheduled flight. The model, based on confidential information from European airlines for 12 types of aircraft ca. 2003, was not transparent with regards to how each of the major components of cost (crew costs, fuel costs, maintenance, depreciation, etc.) impacted that total. This paper describes the development of an airline cost model, based on the Eurocontrol model. The airline cost model explicitly identifies the components of airline costs, is based on US airline cost data, and includes 111 aircraft types. The new model is designed to allow costs to be updated whenever the basic costs change. It considers the type of the aircraft when making calculations, both from the perspective of fuel burn and other costs and uses publically available data on burn rates and block-hour direct operating costs (BHDOC) to obtain these estimates. A case-study analysis of airline costs of operation at 19 major US airports is provided.