“…Macro‐level factors, such as temporal variables [13, 14, 17, 23], seasonal effects [23, 24], airline and airport [13, 14, 25, 27] are widely used as inputs for delay prediction models as described in Table 1. Khanmohammadi et al.…”
This paper presents models for flight delay prediction by considering both the local effects and network effects for the individual airport. Following a complex network approach, the authors analyse the local and network effects separately. Results indicate that the longterm flight delays are mainly caused by network effects, while the short-term flight delays are strongly associated with local delays. Therefore, the existing factors such as temporal variables, weather condition and seasonal effects are replaced with specific novel factors (e.g. crowdedness degree of airport and air traffic system, demand-capacity imbalance) for flight delay prediction. More specifically, this paper shows that the model prediction performance for both classification (predict whether the flight is delayed) and regression (predict the delay values) achieves higher accuracy when using the novel factors. Random Forest algorithms were trained and tested on the U.S. domestic flights in July 2018, and the results show that for classification model, the accuracy, precision and recall score reach 96.48%, 94.39% and 90.26% when classifying delays are within 15 min. Similarly, for regression model, 93.92% of the test errors are within 15 min.
“…Macro‐level factors, such as temporal variables [13, 14, 17, 23], seasonal effects [23, 24], airline and airport [13, 14, 25, 27] are widely used as inputs for delay prediction models as described in Table 1. Khanmohammadi et al.…”
This paper presents models for flight delay prediction by considering both the local effects and network effects for the individual airport. Following a complex network approach, the authors analyse the local and network effects separately. Results indicate that the longterm flight delays are mainly caused by network effects, while the short-term flight delays are strongly associated with local delays. Therefore, the existing factors such as temporal variables, weather condition and seasonal effects are replaced with specific novel factors (e.g. crowdedness degree of airport and air traffic system, demand-capacity imbalance) for flight delay prediction. More specifically, this paper shows that the model prediction performance for both classification (predict whether the flight is delayed) and regression (predict the delay values) achieves higher accuracy when using the novel factors. Random Forest algorithms were trained and tested on the U.S. domestic flights in July 2018, and the results show that for classification model, the accuracy, precision and recall score reach 96.48%, 94.39% and 90.26% when classifying delays are within 15 min. Similarly, for regression model, 93.92% of the test errors are within 15 min.
“…Regarding training and testing, the majority of studies uses traditional scenario (Figure 1.a) [7,18,2]. However, this approach does not consider the possible drifts that are usually present on flight data.…”
Flight delays impose challenges that impact any flight transportation system. Predicting when they are going to occur is an important way to mitigate this issue. However, the behavior of the flight delay system varies through time. This phenomenon is known in predictive analytics as concept drift. This paper investigates the prediction performance of different drift handling strategies in aviation under different scales (models trained from flights related to a single airport or the entire flight system). Specifically, two research questions were proposed and answered: (i) How do drift handling strategies influence the prediction performance of delays? (ii) Do different scales change the results of drift handling strategies? In our analysis, drift handling strategies are relevant, and their impacts vary according to scale and machine learning models used.
“…Previous methods applied to the problem of predicting airplane route delay include the k-Nearest Neighbour algo-rithm (Zonglei, Jiandong, and Guansheng 2008), random forests (Rebollo and Balakrishnan 2014), adaptive networks based on fuzzy inference systems (Khanmohammadi et al 2014), and Markov decision processes incorporating a reinforcement learning strategy (Balakrishna et al 2008). These systems report good performance when the prediction is a single instance that is close in time, but note a concurrent decrease in accuracy as the forecast horizon grows.…”
Flight delays impact airlines, airports and passengers. Delay prediction is crucial during the decision-making process for all players in commercial aviation, and in particular for airlines to meet their on-time performance objectives. Although many machine learning approaches have been experimented with, they fail in (i) predicting delays in minutes with low errors (less than 15 minutes), (ii) being applied to small carriers i.e., low cost companies characterized by a small amount of data. This work presents a Long Short-Term Memory (LSTM) approach to predicting flight delay, modeled as a sequence of flights across multiple airports for a particular aircraft throughout the day. We then suggest a transfer learning approach between heterogeneous feature spaces to train a prediction model for a given smaller airline using the data from another larger airline. Our approach is demonstrated to be robust and accurate for low cost airlines in Europe.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.