Harits Ar Rosyid, Triyanna Widiyaningtyas, Novan Farhandi Hadinata
Food is an important basic requirement because it affects human survival. The availability of food affects the economic stability of a country, one of which results in volatility in food prices. This happens, because of an imbalance between the number of requests and offers. Therefore, a method is needed to identify uncertainties in the pattern of food prices, namely the forecasting method. This study aims to forecast food prices at the provincial level in Java. The stages of research consist of: (1) historical data collection, (2) preprocessing, (3) split datasets, (4) forecasting, (5) validation and evaluation. The dataset is obtained from the National Strategic Food Price Information Center (NSFPIC) page. Testing is done using a comparison of 3 methods, namely Single Exponential Smoothing (SES), Double Exponential Smoothing (DES), and Triple Exponential Smoothing (TES). The optimum parameter validation process uses the Sum of Squared Error (SSE) and error level evaluation process using the Mean Absolute Percentage Error (MAPE). The testing scheme is formed through scenario 1 (5 periods) and scenario 2 (20 periods) to forecast future food price predictions. The testing scheme using scenario 1 obtained an MAPE average of 3.08% and scenario 2 was 8.24%. The level of accuracy produced is very good because it produces an error rate below 10% Scenario 1 gets better results because the forecasting period is shorter, so the causes of factors that influence prices are relatively more constant and create a higher level of data relevance between forecasting data and actual data. © 2019 IEEE.
Universitas Negeri Malang, Electrical Engineering Department, Malang, Indonesia
