|
|
Seven Steps In Forecasting System
1 Determine the use of forecast
2 Select the items to be forecasted
3 Determine the time horizon of the forecast
4 Select the forecasting models
5 Gather the data needed to make the forecast
6 Make the forecast
7 Validate and implement the results
Several realities forecasting faces:
1 seldom perfect
2 assume some underlying stablity
3 family and aggregated forecasted are more accurate than individual product forecast
Quantitative forecasts: Forecasts that employ one or more mathematical models that rely on historical data and/or casual variables to forecast demand.
Qualitative forecasts: Forecasts that incorporate such factors as the decision maker's intution, emotions, personal experiences, and value system.
Overview of Qualitative Method
1 Jury of executive opinion: A forecasting technique that takes the opinion of a small group of high-level managers and results in a group estimate of demand.
2 Sales fore composite: A forecasting technique based on salesperson's estimates of expected sales.
3 Delphi method: A forecasting technique using a group process that allows experts(5-10) to make forecasts.
4 Consumer market survey: A forecasting method that solicits input from customers or potential customers regarding future purchasing plans.
Overview of Quantitative Methods:
1 Time series Models: A forecasting technique that uses a series of past data points to make a forecast.
A time series typically has fore components: trend, seasonality, cycles,random
*Naive approach: A forecasting technique that assumes demand in the next period is equal to demand oin the most recent period.
*Moving average: A forecasting method that uses an average of the n most recent periods of data to forecast the next period.disadvantage less sensitive to real changes/can not pick up trend very well/requiere extensive records of past data)
*Exponential smoothing(二次平滑）: A weighted moving-average forecasting technique in which data points are weighted by an exponentialfunction.
Ft=[aAt-1]+[a(1-a)At-2]+[a(1-a)^2At-2]+.....a is smoothing constant.
#Mean absolute deviation(MAD): A measure of the overall forecast error for a model: MAD=Sigma|forecast errors|/n
#Mean squared error(MSE): The average of the squared differences between the forecasted and observed values.MSE=Sigma(forecast error)^2/n
Ti=b(Ft-Ft-1)+(1-b)Tt-1........b smoothing constant for the trend
*Trend Projections(回归分析）: A time-series forecasting method that fits a trend line to a series of historical data points and then projects the line into the future for forecasts.
*Seasonal variations in data: Regular upward or downward movements in a time series that tie to recurring events.
A multiplicative seasonal model:
a find the average historical demand each season
b compute the average demand over all months by dividing the total average annual demand by the number of seasons
c compute a seasonal index for each season by dividing that month's actual historical demand
d estimate next year's total annual demand
e divide this estimate of total annual demand by the number of seasons
*Cyclical variations in data.
Cycles: Patterns in the data thaht occur every several years.
2 Associative Models.
*Linear-regression analysis: A straight-line mathematical model to describe the functional relationships between independent and dependet variables.
#Standard error of the estimate: A measure of variability around the regression line- its standard deviation.
#Coefficient of correlation: A measure of the strength of the relationship between two variables.
(一元和多元线性回归分析）
3 Monitoring and controlling forecast
Tracking singal: A measurement of how well the forecasting is predicting actual values.
=RSFE/MAD=Sigma(actual demand in period i- forecast demand in period i)/MAD.
2MADs=89%;3MADs=98%;4MADs=99.9%
Bias: A forecast that is consistently higher or consistently lower than actual values of a time series.
Adaptive smoothing: An approach to exponential smoothing forecasting in which the smoothing constant is automatically changed to keep errors to a minimum.
Focus forecasting: Forecasting that tries a variety of computer models and selects the best one for a particular application. Two principles before: sophisticated forecasting models are not always better than simple ones; there is no single technique thaht should be used for all products or services. |
|
楼主: log96
IP卡
狗仔卡
发表于 2005-6-21 10:19
显身卡
