Week 6 Lecture 4 Forecasting

Question 1

When should the ‘Holt-Winters’ method be used?

Answer 1

When data is trended and seasonal.

Question 2

What are 4 advantages fo exponential smoothing?

Answer 2

• Plausible Assumptions
• Easy to compute (for a computer)
• Can be used by staff with limited mathematical background
• Exponential smoothing work well in practise

Question 3

What are 2 disadvantages of exponential smoothing?

Answer 3

• Difficult to find a starting point

- Exponential smoothing lags behind the actual trend as a result of smoothing.

Question 4

How do we calculate the starting point when using the holt-winters method?

Answer 4

Calculate an average of the first 2 or 3 points

Question 5

What are the 3 methods of calculating the accuracy of forecasting methods?

Answer 5

• Mean absolute deviation.
• Mean Square Error.
• Mean Absolute Percentage Error.

Question 6

What are the characteristics of ‘Mean Absolute deviation’? And how is it calculated?

Answer 6

• Considers the absolute value of errors
• Gives equal weighting to each error.

Sum of errors/ N

Question 7

What are the characteristics of ‘Mean Square Error’? And how is it calculated?

Answer 7

• Mean Square Error penalises larger errors so as to not throw off the average.
• If a system has a large anomaly error the Mean Square Error should be used

error (squared) + error (squared) / N

Question 8

What are the characteristics of “Mean Absolute Percentage Error’? And how is it calculated?

Answer 8

• This method gives an average percentage error for each individual error.

Sum of Percentage errors / N

Question 9

What is the error equal to?

Answer 9

Actual - Forecast

Question 10

When dealing with negatives in accuracy checking methods,how are they recorded?

Answer 10

They’re recorded as positive numbers