Forecasting Flashcards
Forecast
A statement about the future value of a variable of interest, such as demand.
Common Features of Forecasts
Assumes same causal system.
Rarely perferct because of randomness
More accurate for groups vs. individuals
Accuracy decreases as time horizon increases.
Forecasting Approaches
Qualitative Methods
Quantitative Methods
Qualitative Methods
Used when situation is vague and little data exist.
Invovles intuition, experience.
Judgmental Forecasts
Delphi Method
Quantitative Methods
Used when situation is ‘stable’ and historical data exists.
Involves mathematical techniques
Time series forecasts and Associative models
Time Series
Time-ordered sequence of observations taken at regular internvals over a period of time.
Time Series Assumption:
Future will be like the past.
Time Series Behaviors
Trend Seasonality Cycle Irregular Variations Random Variations
Types of Time Series Methods
Naive Method Moving Average Weighted Moving Average Exponential Smoothing Trend Exponential Smoothing With Trend Seasonality
Exponential Smoothing
Current Forecast = Previous forecast + a(Previous Actual - Previous Forecast)
The most recent observations might have the highest predicitive value.
More smooth as alpha is increased.
Picking a Smoothing Constant a
Using judgement or trial and error
Balancing smoothness and responsiveness
Low a when stable
High a when susceptible to change
Techniques for Trend
Linear Trend
Nonlinear Trend
Seasonality
Holidays, Weather, Manufacturing year, Fashion year, academic year, sports year
Expressed as variation from average or trend line.
Models of Seasonality
Additive Model
Multiplication Model
Additive Model
Seasonality factor is expressed as a quantity. Simply add or subtract from the series average.
Multiplicative Model
Seasonality is expressed as a percentage of the average amount
Seasonal Relative
Amount by which overall average is multiplied to generate forecast for this season.
Deseasonalize
Historical observations to get nonseasonal components.
Associative Forecasting
Rely on identification of related variable that can be used to predict of the variable of interest.
Associative Techniques
Predictor Variables
Regression
Predictor Variable
Used to predict values of variable interest
Regression
Technique for fitting a line to a set of points.
Choosing a Forecasting Technique
Cost and Accuracy
Short-Term Techniques
Long-Term Techniques
Short-Term Techniques
Moving Average
Exponential Smoothing
Long-Term Techniques
Trend
Delphi
Good indicator of Economy
Sales of Semiconductors
Error:
Difference between the actual value and the value that was predicted for a given period.
Types of Measures of Forecast Accuracy:
Mean Absolute Deviation (MAD)
Mean Square Error (MSE)
Mean Absolute Percentage Error (MAPE)
To find the Seasonal Relative
Find Length
Find the average for the length
If odd, use middle number & divide by average.
Take the seasonal relatives for the time asked for and average them.
Which term most clearly relate to associative forecasing techniques:
Predictor Variables
A simple moving average assigns equal weight to each data point that is represented by the average.
True
Removing the seasonal component from a data series (deseasonalizing) can be accomplished by dividing each data point by its appropriate seasonal relative.
True
Accuracy in forecasting can be measured by
MSE
MAPE
MAD
In trend-adjusted exponential smoothing, the trend adjusted forecast consists of:
An exponentially smoothed forecast and a smoothed trend factor.