1) Which error is more useful for measure accuracy of one method in different time series average error?
A) MAE
B) MAPE
C) RMSE
D) Average error
2) Which error indicate bias (or average over- or under-predicting)?
Average error
A) MAE
B) MAPE
C) RMSE
D) Average error
3) For this set of errors: -1, -4, 0, +2, +3, MAE is:
A) 1.0
B) 1.6
C) 2.0
D) 2.5
4) For this set of errors: -1, -4, 0, +2, +3, RMSE is:
A) Approx. 2.45
B) Approx. 5.48
C) Approx. 1.09
D) 6.0
5) Which of the following is not one of the typical data partitions used in creating a forecasting model?
A) Training
B) Validation
C) Future
D) Trend
6) Before implementing a forecasting model is it important to
A) Rerun the model on the entire series (training+validation) in order to utilize the most recent data in generating your forecasts
B) Rerun the model using the validation period only in order to avoid using ‘old’ data.
C) Make sure that you have found the model with the largest MAE.
D) Utilize an alternative source of data to generate the actual forecasts.
7) Using the “forecast” package in R, what does the “accuracy” function do? (e.g. accuracy(naive.pred, valid.ts) )
A) Checks the accuracy of the R code that you have entered
B) Verifies the accuracy of the data set you are using
C) Generates a set of graphs comparing the forecasts
D) Generates accuracy measures such as MAE, RMSE and MAPE of a forecast
8) What is the primary advantage of looking at prediction cone graphs (like figure 3.6)?
A) They show how MSE changes with different data sets
B) They allow easy comparison of forecast uncertainty between different models
C) They are impressive looking graphs to show off your data visualization skills
D) They can be used to compare only 1-period ahead forecasts