Investigation on the Impact of Attention Mechanism in Deep Learning Models for Temperature Prediction

Abstract

Prediction of the meteorological parameters, such as temperature, humidity, rainfall, wind speed, etc., is a crucial task for industrial and agricultural applications. In recent years deep learning techniques have become more popular for predicting the time series weather data because of their accuracy and promising result. However, adding an attention mechanism in the deep learning model provides more long-term prediction accuracy. This article investigates the potential of attention-based deep learning models for improving the forecasting accuracy of the meteorological parameter temperature. The attention mechanism helps in improving the forecasting accuracy, which is evident from the experimental result analysis in terms of key performance metrics such as Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE).