## 3.7 Exercises

For the following series, find an appropriate Box-Cox transformation in order to stabilise the variance.

`usnetelec`

`usgdp`

`mcopper`

`enplanements`

Why is a Box-Cox transformation unhelpful for the

`cangas`

data?What Box-Cox transformation would you select for your retail data (from Exercise 3 in Section 2.10)?

For each of the following series, make a graph of the data. If transforming seems appropriate, do so and describe the effect.

`dole`

,`usdeaths`

,`bricksq`

.Calculate the residuals from a seasonal naïve forecast applied to the quarterly Australian beer production data from 1992. The following code will help.

`<- window(ausbeer, start=1992) beer <- snaive(beer) fc autoplot(fc) <- residuals(fc) res autoplot(res)`

Test if the residuals are white noise and normally distributed.

`checkresiduals(fc)`

What do you conclude?

Repeat the exercise for the

`WWWusage`

and`bricksq`

data. Use whichever of`naive()`

or`snaive()`

is more appropriate in each case.Are the following statements true or false? Explain your answer.

- Good forecast methods should have normally distributed residuals.
- A model with small residuals will give good forecasts.
- The best measure of forecast accuracy is MAPE.
- If your model doesn’t forecast well, you should make it more complicated.
- Always choose the model with the best forecast accuracy as measured on the test set.

For your retail time series (from Exercise 3 in Section 2.10):

Split the data into two parts using

`<- window(myts, end=c(2010,12)) myts.train <- window(myts, start=2011) myts.test`

Check that your data have been split appropriately by producing the following plot.

`autoplot(myts) + autolayer(myts.train, series="Training") + autolayer(myts.test, series="Test")`

Calculate forecasts using

`snaive`

applied to`myts.train`

.`<- snaive(myts.train) fc`

Compare the accuracy of your forecasts against the actual values stored in

`myts.test`

.`accuracy(fc,myts.test)`

Check the residuals.

`checkresiduals(fc)`

Do the residuals appear to be uncorrelated and normally distributed?

How sensitive are the accuracy measures to the training/test split?

`visnights`

contains quarterly visitor nights (in millions) from 1998 to 2016 for twenty regions of Australia.Use

`window()`

to create three training sets for`visnights[,"QLDMetro"],`

omitting the last 1, 2 and 3 years; call these train1, train2, and train3, respectively. For example`train1 <- window(visnights[, "QLDMetro"], end = c(2015, 4))`

.Compute one year of forecasts for each training set using the

`snaive()`

method. Call these`fc1`

,`fc2`

and`fc3`

, respectively.Use

`accuracy()`

to compare the MAPE over the three test sets. Comment on these.

Use the Dow Jones index (data set

`dowjones`

) to do the following:- Produce a time plot of the series.
- Produce forecasts using the drift method and plot them.
- Show that the forecasts are identical to extending the line drawn between the first and last observations.
- Try using some of the other benchmark functions to forecast the same data set. Which do you think is best? Why?

Consider the daily closing IBM stock prices (data set

`ibmclose`

).- Produce some plots of the data in order to become familiar with it.
- Split the data into a training set of 300 observations and a test set of 69 observations.
- Try using various benchmark methods to forecast the training set and compare the results on the test set. Which method did best?
- Check the residuals of your preferred method. Do they resemble white noise?

Consider the sales of new one-family houses in the USA, Jan 1973 – Nov 1995 (data set

`hsales`

).- Produce some plots of the data in order to become familiar with it.
- Split the
`hsales`

data set into a training set and a test set, where the test set is the last two years of data. - Try using various benchmark methods to forecast the training set and compare the results on the test set. Which method did best?
- Check the residuals of your preferred method. Do they resemble white noise?