nickpoison / tsa4

R code for Time Series Analysis and Its Applications, Ed 4

✨ Here is A Road Map if you want a broad view of what is available.

tsa4

• All the (updated) code used in the text is in textRcode.md

• This is the new home of Time Series Analysis and Its Applications, Edition 4 but the old entry page is still there.

astsa

• The home of astsa - the package for the text is here too.

• See the NEWS for further details about the state of the package and the changelog.

• A demonstration of the capabilities of astsa can be found here at FUN WITH ASTSA

python

• The code in the first 3 chapters of the text has been converted to Python here.

• And a Python package that contains datasets from astsa is here.

• Why use Python if you can use R?     -Mr Natural

   

tsa4's Issues

Example 5.12 of Chapter 5: Additional Time Domain Topics in tsa4

Dear Nicky Poison,

In the Example 5.12 of Chapter 5: Additional Time Domain Topics
https://github.com/nickpoison/tsa4/blob/master/textRcode.md#chapter-5

The R code command line below:

acf(innov)

has created the error as below.

Error in na.fail.default(as.ts(x)) : missing values in object

I guess the arugment "na.action = na.fail" as default value by stats::acf function may be the cause of the error.

So, I set the arugment "na.action = na.pass" in the command line above as

acf(innov, na.action = na.pass)

to have a display of 3 x 3 plots of acf() as result.

I may be wrong.

Please consider the suggestion in your side to check whether my suggestion is wrong or not.

Thanks in advance.

Best regards,
Jimmy Au

tsplot(innov)
head(innov) # contains missing values
u1 u2 u3
1 NA NA NA
2 NA NA NA
3 -7.986554 -3.401042 -6.244936
4 -3.757739 5.160169 -2.446715
5 -2.455558 6.115729 11.073832
6 -1.188275 -2.249943 -10.914951
#acf(innov) # <-------- Error
acf(innov, na.action = na.pass)
acf(innov)
Error in na.fail.default(as.ts(x)) : missing values in object

Wrong ACF code in Example 3.11

The problem

The following code of Example 3.11 is inconsistent with the chart shown in left-hand-side of Fig 3.5.

tsa4/textRcode.md

Lines 567 to 569 in 775a7ec

	ACF = ARMAacf(ar=c(1.5,-.75), ma=0, 50)
	tsplot(ACF, type="h", xlab="lag")
	abline(h=0, col=8)

ACF is a list with value of ACF[1] being actually $\rho(0)=1$. That is, the lag values skewed.

The following figure shows what will be plotted using the code above.

The fix

Actually ACF chart should be start with index 0. Or, if we just ignore ACF[1] (which is always 1), the code should be:

ACF = ARMAacf(ar=c(1.5,-.75), ma=0, 25)[-1]
plot(ACF, type="h", xlab="lag")
abline(h=0)

which produces

The chart above is exactly Fig 3.5 in the text.

R version

> version
               _                           
platform       x86_64-apple-darwin17.0     
arch           x86_64                      
os             darwin17.0                  
system         x86_64, darwin17.0          
status                                     
major          4                           
minor          0.2                         
year           2020                        
month          06                          
day            22                          
svn rev        78730                       
language       R                           
version.string R version 4.0.2 (2020-06-22)
nickname       Taking Off Again