Dear SPlus users,
I am dealing with some research in time series. At this stage I need to use
two tests; this is my first time, I have never used tests before.
Altough I have got a lot of references and books, I am not able to understand
completely how to perform these tests and what can be already found in Splus.
Could somebody help me in these first steps?
I am using Splus4
thank you for your attention and your eventual help
Stefano
QUESTION 1: ABOUT THE MANN-KENDALL NON-PARAMETRIC TEST
I need to use the Mann-Kendall rank test to determine whether the trend
applied to a time series is significant or not.
I tried to use the cor.test command, but I am not sure I did the right thing.
I put with
the original time series called mydata as x
the trend called mytrend as y:
cor.test(mydata, mytrend, alternative="two.sided", method="kendall")
and this is the output:
Kendall's rank correlation tau
data: mydata and mytrend
normal-z = 6.6512, p-value = 0
alternative hypothesis: true tau is not equal to 0
sample estimates:
tau
0.2864674
Is what I did correct?
What does it mean? Is this informations complete and exaustive?
QUESTION 2: ABOUT THE KOLMOGOROV SMIRNOV TEST
I need to use the Lolmogorov Smirnov test in order to test whether a fitted
Gamma distribution and observed data are from the same population.
In particular
observed data are a time series called mydata
and the fitted Gamma distribution is a Gamma of shape = 12.66 and rate = 1.41
In Splus the command is ks.gof:
ks.gof(x, y = NULL, alternative = "two.sided", distribution = "normal", ...)
I suppose I have to insert
mydata as x
alternative = two.sided
distribution = gamma
what about y?
And then? I should be able to detrmine a critical value. What is it for? And
how can I choose it, if I can?
Is this Splus command exaustive for the Lolmogorov Smirnov test?
thank you again
|