Analysis of the Backpropagation Algorithm in Viewing Import Value Development Levels Based on Main Country of Origin

Abstract

In this study, we will discuss the development of import values in general based on the main country of origin in units of net weight (tons). The import value to be discussed is focused on the value of imports in North Sumatra Province, whose data is obtained from the Central Bureau of Statistics in 2009-2014. The algorithm that will be used to see and measure the level of development of import values is the backpropagation algorithm. This algorithm is one part of the Artificial Neural Network algorithm that is quite reliable in solving problems related to times series data. With the backpropagation algorithm, 5 architectural models will be formed that will be used to measure the level of development of import values, namely 4-12-1 architectural models, 4-15-1 models, 4-18-1 models, 4-19-1 models and model 4-20-1. From the 5 models, one of the best that will be used as a reference to measure the development of import values will be chosen. After analyzing the best models were obtained 14-19-1 with the level of truth produced more than 90%. With this architectural model, calculations will be made on the level of development of import values based on the main country of origin for 2015 to 2020.