Implementation of Convolutional Neural Network and Multilayer Perceptron in Predicting Air Temperature in Padang

Isman Kurniawan,Lusi Sofiana Silaban,Devi Munandar

doi:10.29207/resti.v4i6.2456

Abstract

Weather prediction is usually performed for a reference in planning future activity. The prediction is performed by considering several parameters, such as temperature, air pressure, humidity, wind, rainfall, and others. In this study, the temperature, as one of weather parameters, is predicted by using time series from January 2015 to December 2017. The data was obtained from Lembaga Ilmu Pengetahuan Indonesia (LIPI) weather measurement station in Muaro Anai, Padang. The predictions were carried out by using Convolutional Neural Network (CNN), Multilayer Perceptron (MLP), and the hybrid of CNN-MLP methods. The parameters used in the CNN method, such as the number of filters and kernel size, and used in the MLP method, such as the number of hidden layers and number of neurons, were selected by performing the hyperparameter tuning procedure. After obtaining the best parameters for both methods, the performance of both methods was evaluated by calculating the value of Root Mean Square Error (RMSE) and R2. Based on the results, we found that the prediction by CNN is more accurate than other method. This is indicated by the highest value of R2 of the prediction obtained by CNN method.

Highlights

Introduction developed anArtificial Neural Network (ANN) model to predict daily mean ambientAir temperature is one of the weather parameters that have an important effect on daily life
The ability to predict the air temperature accurately is important in planning steps of a certain activity, such as flight recommendation, agriculture, sailing, etc
A numerical approach utilizes the temperatures in Denizli, Turkey, and found that ANN is reliable to be used for the prediction[2]

Summary

Introduction

ANN model to predict daily mean ambient. Air temperature is one of the weather parameters that have an important effect on daily life. An empirical approach is performed by gathering data using observation of soil, satellite, etc. A numerical approach utilizes the temperatures in Denizli, Turkey, and found that ANN is reliable to be used for the prediction[2]. In 2015, Chithra and coworkers performed implemented ANN in a model to predict mean monthly maximum and minimum temperature in Kerala, India[3]. In 2015, Appelhans and coworkers predict monthly air temperature at Mt. Kilimanjaro, Tanzania by using 14 machine learning algorithms. Other studies were performed by implementing the ANN model to predict air temperature[5]–[7]

Objectives

Methods

Results

Conclusion