Week 10 - Forecasting Flashcards
How to workout a 3 week moving average?
Add the sales for the last 3 weeks then divide by three which is the period
Example/explain moving average
Moving average starts at week 4 as it is an average of the previous 3 weeks which goes all the way down to week 13
What is a forecast error?
Is the difference between the sales of a particular week and the moving average of the same week
What is the squared forecast error? And it purpose?
• is the forecast error squared which helps to eliminate the negatives in the forecast error
• This gives us the difference between the actual sales and the forecasted sales
What is the mean squared error (MSE) and root mean squared error (RMSE)
• Mean squared error is the mean of the squared forecast error
• Root MSE is the square root of the squared forecast error
How will a moving average gain stability?
When there is a greater number of periods used in the average
What is the problem with using a moving average over too many periods?
Average will be so stable that it will be slow to respond to non random changes in the demand data
What is responsiveness?
Is the ability of a forecast to adjust quickly to true changes in the base level of demand
What is exponential smoothing?
Is a method which keep a running average of demand and adjusts it for each period in proportion to the difference between the latest actual demand figure and the latest value of the average
Exponential smoothing equation (4)
• SF+1 = smoothed forecast for time period following t
• SFt = smoothed forecast for period t
• α = smoothing constant that determines weight given to previous data
• At = actual demand in period
Characteristics of the smoothing constant (α) (3)
• Is a decimal between 0 and 1 where 0 is most
stable, 1 is most responsive (values between 0.1 and 0.3 are often used in practise)
• A larger value of α gives very little smoothing
• A smaller value gives considerable smoothing/damping
What should the Smoothing constant (α) be for a time series with little random variability?
Larger values of the smoothing constant (α) are better as they react quicker to changes
What should the Smoothing constant (α) be for a time series with Large random variability?
Lower values of α are best so they do not overreact and adjust the forecasts too quickly
Exponential smoothing forecast
Only need sales figure for the previous week and current then get the average
Benefits of smoothing methods (3)
• Simple and low cost
• More accuracy may be obtained with more sophisticated or decomposition methods
• Are appropriate for forecasts of thousands of items in inventory systems
