6SIGMA工具_ARIMAForecasting.pptVIP

Breakthrough Management Group Breakthrough Management Group Version 1.0 INTRODUCTION TO ARIMA FORECASTING MODELS Forecasting Topics Forecasting Introduction Stationary Time Series The “I” of ARIMA Autoregressive Model (AR) Identification, Estimation and Diagnosis, Forecasting Moving Average Model (MA) Identification, Estimation and Diagnosis, Forecasting Autoregressive Moving Average Model (ARMA) Identification, Estimation and Diagnosis, Forecasting ARMA Seasonal Models Identification Case Studies Non Seasonal - Shipping Backlog Seasonal - Emergency Room Admission Backshift Operator Forecasting - Introduction Forecasting is a common activity in the operation of most enterprises and institutions. Anyone in a position of planning, controlling, and managing projects, people, finance, and process operations has an ongoing need for estimates of the future Forecasts may form the basis of staffing decisions, process modifications, and budgeting and finance projections Forecasting - Introduction In most situations, the forecasts are for processes and activities that have a quantifiable historical record called a Time Series. Time series have the following characteristics: The time periods are of equal length Data is arranged from earliest period to latest period The process or activity is consistent over time The method of measurement remains consistent over time Forecasting - Introduction A monthly Sales Time Series is Tabled below. The easiest and most convenient way to see the behavior of a time series is to graph it. Open Forecast1.mtw StatTime SeriesTime Series Plot Graph: C2 Date/Times Stamp: C1 Jan 99 13727 Feb 99 16109 Mar 99 15589 Apr 99 14741 May 99 17205 Jun 99 16373 Jul 99 15286 Aug 99 17364 Sep 99 16376 Oct 99 16225 Nov 99 18600 Dec 99 17570 Forecasting - Introduction The objectives of time series analysis are to: Obtain a concise description of the time series Construct a model to explain time series behavior Test the model against histo