Open AccessHybrid vector autoregression–recurrent neural networks to forecast multivariate time series jet fuel transaction price
Author(s) -
Agung Bayu Aji,
Isti Surjandari
Publication year - 2020
Publication title -
iop conference series. materials science and engineering
Language(s) - English
Resource type - Journals
eISSN - 1757-899X
pISSN - 1757-8981
DOI - 10.1088/1757-899x/909/1/012079
Subject(s) - jet fuel , vector autoregression , multivariate statistics , computer science , artificial neural network , aviation , transaction cost , database transaction , recurrent neural network , econometrics , economics , artificial intelligence , engineering , machine learning , finance , aerospace engineering , database
Fuel cost is the most contributed component in the operational cost of all transportation modes. In the aviation industry, jet fuel cost contributed to a percentage of 33.33% of the total airline operational costs. To increase efficiency in operational costs and the airline should have jet fuel price monitoring systems that can forecast the future price and give some strategy recommendations to airlines. In this research, we propose many multivariate time series-based predictive analytics as a tool for the airline to monitor and forecast the jet fuel price transaction based on jet fuel transaction price. We consider the global crude oil price and also global and local jet fuel prices in each airport. We also consider additional variables for the economical aspect that applied differently for each airport location. We examine two Recurrent Neural Network (RNN) algorithm, Long Short Term Memory (LSTM) and Gate Recurrent Units (GRU). For minimizing the weakness of LSTM and GRU, we combine each methods with Vector Autoregression (VAR). After forecasting results using VAR-LSTM and VAR-GRU, we get forecasting accuracy of 98.98% and 99.40% respectively.