FORECASTING USING FUZZY LOGIC RELATIONSHIP GROUPS METHOD ON IMMIGRATION PUBLIC SERVICE’S PASSPORT PRODUCTION DATA

Abstract

Stationarity is a term used to describe the pattern of trend in time series data. In time series data, this term known as stationary and non-stationary. Non-stationary data is a data that has an unstable pattern of increase and decrease. This condition makes forecasting more difficult. Fuzzy Time Series is one of many forecasting methods that can be used. In this algorithm, adding order is an option that can be used to increase the accuracy of the method. Application up to order three are carried out to determine the effect of addition order to the resulted accuracy value. Experiment is done by applying the used method to the data which is divided into several amounts of data. From the experiment, the average accuracy value of the three Order of Fuzzy Logic Relationship Groups (FLRG) Order-1, Order-2, and Order-3 are 84.06719%, 85.77546%, 92.01034%. FLRG Order-3 has the largest accuracy value while the smallest accuracy value is owned by FLRG Order-1. From this, it is proven that the addition of order able to reduce the error in accuracy value while forecasting using non-stationary data but the accuracy produced by different amounts of data are erratically increasing and decreasing. the experiment concluded that the order, the amount of data, and the data pattern are factors that affect the accuracy result.