PERAMALAN PRODUKSI DAN LUAS AREA TANAM TEMBAKAU DI INDONESIA DENGAN METODE ARTIFICIAL NEURAL NETWORK

Abstract

Tobacco is a leading commodity in the plantation sector which is very important in the Indonesian
economy. Products that can be produced or traded are tobacco leaves and cigarettes. In 2012, the world's
largest tobacco producing countries were China, Brazil, China, and the United States. Indonesia ranks 5th as
the world's largest tobacco producer with a total production of 226,704 tons or 3.0% of total global tobacco
production. Indonesia's success in producing tobacco can be determined by forecasting the next few years.
The data used are production data and the area of tobacco planted in Indonesia from 1975 - 2018. The method
used is Backpropagation which is a supervised training method applied to the Artificial Neural Network. One
of the algorithms from Backpropagation is the Levenberg Marquardt algorithm developed by Donald
Marquardt and Kenneth Marquardt, which provides a solution to the problem of minimizing non-linear
functions. In this study it can be concluded that the best model for tobacco production is 1-6-1 with an MSE
training value of 0.153 so that the forecast in 2019 is 160,240 tons. Meanwhile, the best model for tobacco
planting area is 1-16-1 with MSE training value of 0.102, so the forecast in 2019 is 190,000 Ha.
Keywords: Levenberg Marquardt Algorithm, Artificial Neural Network Method, Tobacco Production