4QA3 F12 Week 5 Lecture Notes

85 3930 4113 4002 4302 4731 4636 4738 5234 5336

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: appropriate for stationary series o  Both methods depend on a single parameter o  Both methods lag behind a trend o  One can achieve the same distribution of forecast error by setting α = 2/(n + 1). ●  Differences o  ES carries all past history. MA eliminates “bad” data after N periods. o  MA requires all n past data points while ES only requires last forecast and last observation. 4QA3 F12 25 ●  We assume the following form for a trend series: ●  Dt = µ + Gt + εt Two common trend- based methods are: Demand Patterns o  Regression, o  Holt’s Method. 4QA3 F12 26 Here, the independent variable corresponds to time: ●  Y 15 Dt = a + bt Regression for Time Series 10 error7 eries 5 error2 • Regression can be used when trend is present Model: Dt = a+bt Observed t is scaled to 1,2,3…then we set • If Y (actual Y) Predicted Y (predicted Y) 0 ecast error by 0 5 10 15 X S xy = n∑ iDi − [ n(n + 1) / 2]∑ Di i ad...
View Full Document

{[ snackBarMessage ]}

Ask a homework question - tutors are online