NURMALASARI, 101011158 (2014) APLIKASI AUTOREGRESSIVE INTEGRATED MOVING AVERAGE (ARIMA) UNTUK PERAMALAN JUMLAH KASUS PERCERAIAN. Skripsi thesis, UNIVERSITAS AIRLANGGA.

Preview	Text (ABSTRAK) ABSTRAK.pdf Download (173kB) | Preview
	Text (FULLTEXT) fulltext.pdf Restricted to Registered users only Download (2MB) | Request a copy

Abstract

Forecasting is an activity to predict an event occurring in the future. Quantitative forecasting technique consists of classical and modern forecasting techniques. Selection of some methods in classical forecasting techniques based on specific data types causing limitations in its use, so most of forecasting activities using ARIMA method, because this method can be used on almost any types of data and has high accuracy. Therefore, the purpose of this study was to apply ARIMA method that used to forecast divorce cases number in the Religious Court, Lumajang. This study used 60 historical data points (January 2009-December 2013). The data were analyzed using Minitab 17 software to get fit model based on ARIMA method. Then, forecasting results were evaluated using 12 historical data points (January-December 2013) using MPE and MAPE to determine the level of forecasting accuracy. The results of this study was talaq divorce number in 2014-2015 can be forecasted using ARIMA (3,1,1) equation with (1 - B)(1 + 0,7874B + 0,7401B2 + 0,5282B3)Xt =(1-0,9911B)et. The forecasting results of talaq divorce with ARIMA (3,1,1) indicated there were 1073,5550 cases in 2014 and 1093,7091 cases in 2015 with forecast's degree of accuracy was 0,943% (MPE) and 13,596% (MAPE). While contested divorce number in 2014-2015 can not be predicted using ARIMA method. ARIMA (2,1,0) which has been formed was declared unfit for forecasting model because ARIMA model residuals are not white noise (not random). The conclusion of this study was ARIMA method can be used to forecast talaq divorce number for 2014-2015 in the Religious Court, Lumajang, with a high percentage of accuracy. To overcome the ARIMA model residuals were not white noise (not random), such as contested divorce data, can use GARCH methods or other methods of time series forecasting.

Actions (login required)

View Item