Temperature forecasting via neural networks based on time-series data
No Thumbnail Available
Date
2021
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Oum El Bouaghi
Abstract
in this paper, we presentdeep neural network architectures and use themintime series weather forecasting. We started with a MLP (Multi- Layer Perceptron) architecture, then we used CNN (Convolutional Neural Network) architecture, and moving on to LSTM (Long-Short Term Memory),
to finally making a hybrid architecture of CNN and LSTM. Eight years (2009-2016) of hourly meteorological data were used to train each model
and the results show that neural network could be a good way to predict future temperatures.
Description
Keywords
Artificial neural networks ; Temperature forecasting ; Time series ; MLP; CNN ;
LSTM ; Time series forecasting