Presenter Email
desmond.bisandu@cranfield.ac.uk
Other Topic Area
Flight delay classification and prediction
Keywords
Analysis, BiLSTM, deep learning, Flight delay, Machine learning, Modelling
Abstract
Flight delays can be prevented by providing a reference point from an accurate prediction model because predicting flight delays is a problem with a specific space. Only a few algorithms consider predicted classes' mutual correlation during flight delay classification or prediction modelling tasks. None of these existing methods works for all scenarios. Therefore, the need to investigate the performance of more models in solving the problem of flight delay is vast and rapidly increasing. This paper presents the development and evaluation of LSTM and BiLSTM models by comparing them for a flight delay prediction. The LSTM does the feature extraction in both models, except that the BiLSTM maintains an equilibrium with a forward and backward hidden sequences model training. The experimental results show that the BiLSTM model accuracy improved by 97.56%, with a 21.11% accuracy increase. Furthermore, the models' performance results and confusion matrix shows how the BiLSTM model outperforms the LSTM model. In evaluating the MCC, the BiLSTM model offers a better mutual correlation among the predicted classes with 0.9944. Our findings suggest that for predicting flight delays, the BiLSTM model utilises the advantages of the bidirectional hidden sequences and the deep neural network for exploitation and exploration of best performance given a high accuracy, precision, recall and F1-Score results. Hence, we can recommend the BiLSTM in developing a decision support system for flight delays and related applications.
Included in
Aerospace Engineering Commons, Aviation Commons, Computer Sciences Commons, Data Science Commons
A bidirectional deep LSTM machine learning method for flight delay modelling and analysis
Flight delays can be prevented by providing a reference point from an accurate prediction model because predicting flight delays is a problem with a specific space. Only a few algorithms consider predicted classes' mutual correlation during flight delay classification or prediction modelling tasks. None of these existing methods works for all scenarios. Therefore, the need to investigate the performance of more models in solving the problem of flight delay is vast and rapidly increasing. This paper presents the development and evaluation of LSTM and BiLSTM models by comparing them for a flight delay prediction. The LSTM does the feature extraction in both models, except that the BiLSTM maintains an equilibrium with a forward and backward hidden sequences model training. The experimental results show that the BiLSTM model accuracy improved by 97.56%, with a 21.11% accuracy increase. Furthermore, the models' performance results and confusion matrix shows how the BiLSTM model outperforms the LSTM model. In evaluating the MCC, the BiLSTM model offers a better mutual correlation among the predicted classes with 0.9944. Our findings suggest that for predicting flight delays, the BiLSTM model utilises the advantages of the bidirectional hidden sequences and the deep neural network for exploitation and exploration of best performance given a high accuracy, precision, recall and F1-Score results. Hence, we can recommend the BiLSTM in developing a decision support system for flight delays and related applications.
Comments
Presented in Session 3 A - Advancing Aviation: AI & Machine Learning