Rice Price Prediction Using Multilayer Perceptron (MLP) and Long Short Term Memory (LSTM).

  • Steven Sen Universitas Trisakti
  • Dedy Sugiarto
  • Abdul Rochman

Abstract

Rice is one of the main commodities in Indonesian society. The main problem with rice nationally is inflation of rice prices. Therefore, this research predicts the price of rice using Multilayer Perceptron (MLP) artificial neural network architecture and deep learning: Long Short Term Memory (LSTM) to anticipate these problems. The data used in this study are real data on rice prices during 2016 - 2019 obtained from PT. Food Station. The total dataset is 1307 with the distribution of 1123 as data train and 184 as test data. The final results obtained in this study are LSTM superior to MLP, with the value of Root Mean Square Error (RMSE) training data: 0.49 RMSE loss value of test data is 0.27. The most optimal LSTM model from 3 tests was carried out, namely the number of hidden layers = 16 and epochs = 150 times.

Downloads

Download data is not yet available.
Published
2020-07-02
How to Cite
Sen, S., Sugiarto, D., & Rochman, A. (2020). Rice Price Prediction Using Multilayer Perceptron (MLP) and Long Short Term Memory (LSTM). Ultimatics : Jurnal Teknik Informatika, 12(1), 35-41. https://doi.org/https://doi.org/10.31937/ti.v12i1.1572
Section
Articles