ANALISA KOMPARASI NEURAL NETWORK BACKPROPAGATION DAN MULTIPLE LINEAR REGRESSION UNTUK PERAMALAN TINGKAT INFLASI
ABSTRACT: The inflation rate
can not be underestimated in a country's economic system and businesses in
general. If inflation can be predicted with high accuracy, of course, can be
used as the basis of government policy making in anticipation of future
economic activity. In this study will be used back propagation neural network
method and multiple linear regression method to predict the monthly inflation
rate in Indonesia, then compare which method is the better. The data used comes
from the central statistical agency in 2006-2015, which is 80% as training data
and 20% as testing data. In the results of the data analysis is concluded that
the performance of multiple linear regression is better than back propagatin
neural network, with a mean absolute deviation (MAD) is 0.0380, a mean square
error (MSE) is 0.0023, and a Root Mean
Square Error (RMSE) is 0.0481.
Penulis: Amrin AIN
Kode Jurnal: jptkomputerdd160294
