DATA PATTERN ANALYSIS AS AN ALTERNATIVE

TO DETERMINE THE ORDER OF INTERVENTION MULTI INPUTS MODEL

Dewi Anugeraheni Sahari

Abstract. Intervention model is a model in time series which is used to explore the impact on the series from external factors which gives an estimate to the observed variables (Ismail, 2009). In developing intervention models, the accuracy of identifying the order of the interventions is needed. According to Box and Reinsell (2008), the identification may be aided by direct inspection of the data to suggest the form of effect due to the known event, and supplementary evidence may sometimes be available from examination of the reseals from a model fitted before the intervention term is introduced. Ismail et. al (2009) and Nuvitasari et. al (2008) studied more about the determination of the intervention order  through the residual model. In this research, it will analyze the data pattern to determine the orders of the intervention. To determine intervention orders through data pattern, it must be known about the characteristics of intervention order. To get the characteristics, some of response pattern with some intervention orders will be simulated.  Furthermore, the intervention orders can be determined by adjusting the data pattern with the characteristics pattern obtained.

The result of this study is applied to CPI (Consumer Price Index) data of Surakarta in January 2000 November 2009 and the national CPI in January 1995 July 1998. In this application, the intervention orders are identified by two methods, through the residual pattern and data pattern. The results showed that the identification of the order of interventions through data pattern is easier and more efficient.

Keyword: Time series, intervention, multiple intervention inputs model, intervention order, data pattern, residual pattern, response pattern.

[read more...]