This preview shows page 1. Sign up to view the full content.
Unformatted text preview: curacy
BIAS The average of forecast errors of the Time
Should to be as close to zero as possible.
“+” indicates under-forecast of the TS
“-” indicates over-forecast of the TS Bias should equal zero if using regression.
Compare regression models using std. error of
significant Weighted Moving Averages
Weighted Types of Time Series Models
Exponential smoothing New forecast is equal to the “old forecast”
plus an adjustment
Specify a smoothing constant (a fraction of
forecasting discrepancy or error)
Between 0 and 1; determines the magnitude of
adjustment made to the “old forecast” Exponential Smoothing
(Start with Actual as Forecast) How are we doing so far?
The Summary Section Types of Time Series Models
Simple Regression Criterion V.: Variable you want to forecast...
View Full Document
This document was uploaded on 02/19/2014.
- Spring '14