Statistical methodological review for time-series data

Abstract

Numerous literatures on statistical methods for Time-Series (TS) data have been published. In this paper, a literature of the TS data analysis methods is reviewed. We organize the review based on the basic three-family category of TS models: The Exponential Smoothing Model (ESM) family, the Auto-Regressive Integrated Moving Average (ARIMA) model family, and the Unobserved Component Model (UCM) family. A roadmap is provided in a diagram format to these TS methods which are translatable into nowadays computing statements. Further, the execution of these methods in SAS commands (as one of the most popular nowadays statistical software packages) is also presented. This paper will be very beneficial for practitioners, forecasters, and researchers in diverse fields of study (such as business, management, finance, economics, etc.) to determine which TS data analysis method (along with the corresponding SAS command) are ready to use.

Subject Classification: (2010):

• 37M10 − Time Series Analysis
• 62M10 – Time series, auto-correlation, regression, etc
• 91B84 − Economic Time Series Analysis

Keywords:

• Auto-correlation
• Forecast
• Time Series
• Exponential Smoothing Model (ESM)
• Auto-Regressive Integrated Moving Average (ARIMA)
• Unobserved Component Model (UCM)