THE EFFECT OF THE NUMBER OF HIDDEN LAYERS IN THE BACKPROPAGATION IN CASE STUDY WEATHER CLASSIFICATION

Abstract

In this modern era, there are many algorithms that can be used to classify the weather, one of this algorithms is Backpropagation. Using Backpropagation, this research were using temperature, pressure, humidity, wind speed, rain and clouds as input parameters. And the output are clear, clouds and rain. Backpropagation consists of learning process and testing process. The learning process is to get optimal weight and testing process is to test the classification using the optimal weight from learning process. Data that was used in this research are 1600 data (80%) for learning process and 400 data (20%) for testing process. This research using Backpropagation with 1, 2 and 3 hidden layer to examine the accuracy result of weather classification. As a result, there is no significant changes of percentage accuracy on Backpropagation with 1, 2 and 3 hidden layers. But, the more number of hidden layer it used, the more epoch it requires.