Crude oil price prediction using temporal fusion transformer model

In this paper, we applied the temporal fusion transformer model to the crude oil price movement modeling and forecasting. The temporal fusion transformer model has been adopted in the crude oil price forecasting model using the attention mechanism, to capture the different level of autocorrelations among observations in the crude oil prices. Empirical evaluation of the transformer based multi-horizon ahead crude oil price forecasting model has been conducted. Experiment results show that the introduction of transformer model in the forecasting process has improved the forecasting accuracy significantly at longer time horizon. © 2023 The Authors. Published by Elsevier B.V.