Professional Documents
Culture Documents
Forecasting technique
Time series analysis
Forecasting errors
Using EXCEL
Forecasting
techniques (pg. 436 Exhibit 11.1)
1. Statistical (Time Series, Causal)
2. Judgement/Qualitative (Expert opinion, Market Survey, Delphi)
An Example
1 650
2 700 0.1*650+0.9*650
=
650
3 810 0.1*700+0.9*650 =
655
783 683.5
852
Illustration
Pg.472 Problem 2
Exercise:
pg473 Problem 9
Forecasting errors
Mean Absolute Deviation (MAD) (pg.448)
An Example
Number of Accuracy
MADs
+/- 1 57%
+/- 2 88.9%
+/- 3 98.3%
+/- 4 99.9%
With 57% accuracy, the forecast demand for July using 3-mth Wt. Moving
Average = 780 +/-
108 (672 to 888)
With 88.9% accuracy, the forecast demand for July using 3-mth Wt. Moving
Average = 780+/-
2*108 (564 to 996)
Exercise:
pg.471 Problem 3, 11
Regression Analysis
Assumptions
1. Linear -- the past data and future projections are fall about a
straight line (least squares
method: minimize the sum of squared forecast
error)
2. Time is the independent variable, x
Y = a + bx
An example
1 31 31 1
2 40 80 4
3 30 90 9
4 34 136 16
5 25 125 25
6 20 120 36
b = (582-6*3.5*30)/(91-6*3.5*3.5) = -2.7
a = 30-(-2.7)*3.5 = 39.6
Y = 39.6 - 2.7x
Time Code
January -2
February -1
March 0
April 1
May 2
January -3
February -1
March 1
April 3
If then
Example
b = -96/70 = -1.4
a = 30
Y = 30 - 1.4x
Forecasting error
= 5.25
or
= 5.25
Number of Accuracy
Syx
+/- 1 68%
+/- 2 95.5%
+/- 3 99.7%
Exercise:
Pg.471 Problems 17
Using Excel
1. Click Tools, Click Data Analysis
2. Choose Moving Average/Exponential Smoothing/Regression
Method Parameter Excel Reminder
terminology