Samuel H. Huang, Winter 2012
Basic Concepts and Constant Process
• Overview of demand forecasting
• Constant process– Average and moving average method
– Exponential Smoothing method
Samuel H. Huang, Winter 2012
Demand Influencing Factors
• Product characteristics• Past demand• Economic condition• Competition• Planned marketing efforts• Planned price discount
Samuel H. Huang, Winter 2012
Forecasting Methods
• Qualitative: rely on human judgment– Market survey (customer response)
– Delphi technique (expert opinion)
• Causal: demand is highly correlated with certain factors
• Time Series: past demand is a good indicator of future demand
• Simulation: mimic consumer behavior to conduct what-if analysis
Samuel H. Huang, Winter 2012
Time-series Forecasting
• Constant process– Average– Moving average– Exponential smoothing
• Trend process– Regression– Double exponential smoothing
• Seasonal process
Samuel H. Huang, Winter 2012
Characteristics of Forecast
• Forecasts are always wrong and thus should include an error analysis
• Long-term forecasts are usually less accurate than short-term forecasts
• Aggregate forecasts are usually more accurate than disaggregate forecasts
Samuel H. Huang, Winter 2012
Constant Model: Average
• Constant model
• Forecast
• Derived based on minimizing the sum of squared errors
tt ad
N
da
Nt t 1ˆ
ade tt ˆ
0ˆ2ˆ
ˆ
ˆ 11
21
2
N
t t
Nt t
Nt t ad
ad
add
ad
ed
Nt t aNd1 ˆ
Samuel H. Huang, Winter 2012
Moving Average
• Average only the most recent data points
• Smooth out noise
• Can respond to change in process
n
dM
tnt t
t
1
n
ddMM ntt
tt11
1
Samuel H. Huang, Winter 2012
Exponential Smoothing
• Adjust forecast based on the most recent data point
• It is a weighted average of all historical data points, with the weight decreasing exponentially with the age of the data point
• Different initial estimates can be used – average of several past data points
11 ttt SdS
Samuel H. Huang, Winter 2012
Example: Exponential Smoothing
Microsoft Office Excel 97-2003 Worksheet