The Covid-19 pandemic that has hit the world, especially Indonesia, has greatly disturbed the stability of the inflation rate. Inflation that continues to increase will disrupt the economy in this country. Therefore this study aims to analyze the ability of ANN backpropagation which will be applied to predict the development of the inflation in Indonesia during the Covid-19 pandemic so that later it can be useful information for the government and society. The research data used is inflation data according to expenditure groups obtained from CBS (Central Statistics Agency) in January-May 2020. Prediction is done using the backpropagation neural network algorithm. This paper uses four network architectures, namely: 3-5-1, 3-10-1, 3-25-1 and 3-50-1. Based on the training and testing of the four models, the 3-10-1 model is the best architectural model that is suitable for predicting the development of the inflation in Indonesia with an accuracy of 75%. Also, this model performs an iteration of 25303 and an MSE test of 0.0362820326. Based on the prediction results in June-August 2020 and real data obtained from the Central Statistics Agency, ANN using the backpropagation method is highly recommended to be used to predict the development of Indonesian Inflation according to the Expenditure Group.
Copyrights © 2021