R help - Dec 2018 - ks.test ; impossible to calculate exact exact value with ex-aequos

Dear all,

I'm trying to use ks.test in order to compare two curve. I've 0 values i
think this is why I have the follonwing warnings :impossible to calculate
exact exact value with  ex-aequos


a=c(3.02040816326531, 7.95918367346939, 10.6162790697674, 4.64150943396226,
1.86538461538462, 1.125, 1.01020408163265, 1.2093023255814,
0.292452830188679,
0, 0, 0)
b=c(2.30769230769231, 4.19252873563218, 5.81924882629108, 6.2248243559719,
5.02682926829268, 4.50728862973761, 3.61741424802111, 5.05479452054795,
3.68095238095238, 1.875, 5.25, 0)

ks.test(a,b)

data:  a and b
D = 0.58333, p-value = 0.0337
alternative hypothesis: two-sided

Warning message:
In ks.test(a, b) :
impossible to calculate exact exact value with ex-aequos

Does the p-value is correct ? Otherwise, how to solve this issue ?

Thanks a lot.

	[[alternative HTML version deleted]]

Hello,

That is a warning, not an error.
And it documented. In ?ks.test, section Details, the relevant part is


The presence of ties always generates a warning, since continuous 
distributions do not generate them. If the ties arose from rounding the 
tests may be approximately valid, but even modest amounts of rounding 
can have a significant effect on the calculated statistic.


It then adds, three paragraphs down:

Exact p-values are not available for the two-sample case if one-sided or 
in the presence of ties.


You can use exact = FALSE but it will still issue a warning.

ks.test(a, b, exact = FALSE)

	Two-sample Kolmogorov-Smirnov test

data:  a and b
D = 0.58333, p-value = 0.0337
alternative hypothesis: two-sided

Warning message:
In ks.test(a, b, exact = FALSE) :
   p-value will be approximate in the presence of ties



Hope this helps,

Rui Barradas
?s 21:17 de 10/12/2018, Fatma Ell escreveu:
> Dear all,
> 
> I'm trying to use ks.test in order to compare two curve. I've 0
values i
> think this is why I have the follonwing warnings :impossible to calculate
> exact exact value with  ex-aequos
> 
> 
> a=c(3.02040816326531, 7.95918367346939, 10.6162790697674, 4.64150943396226,
> 1.86538461538462, 1.125, 1.01020408163265, 1.2093023255814,
> 0.292452830188679,
> 0, 0, 0)
> b=c(2.30769230769231, 4.19252873563218, 5.81924882629108, 6.2248243559719,
> 5.02682926829268, 4.50728862973761, 3.61741424802111, 5.05479452054795,
> 3.68095238095238, 1.875, 5.25, 0)
> 
> ks.test(a,b)
> 
> data:  a and b
> D = 0.58333, p-value = 0.0337
> alternative hypothesis: two-sided
> 
> Warning message:
> In ks.test(a, b) :
> impossible to calculate exact exact value with ex-aequos
> 
> Does the p-value is correct ? Otherwise, how to solve this issue ?
> 
> Thanks a lot.
> 
> 	[[alternative HTML version deleted]]
> 
> ______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>

On Mon, 2018-12-10 at 22:17 +0100, Fatma Ell wrote:
> Dear all,
> I'm trying to use ks.test in order to compare two curve. I've 0
values i
> think this is why I have the follonwing warnings :impossible to calculate
> exact exact value with  ex-aequos
> 
> a=c(3.02040816326531, 7.95918367346939, 10.6162790697674, 4.64150943396226,
> 1.86538461538462, 1.125, 1.01020408163265, 1.2093023255814,
> 0.292452830188679,
> 0, 0, 0)
> b=c(2.30769230769231, 4.19252873563218, 5.81924882629108, 6.2248243559719,
> 5.02682926829268, 4.50728862973761, 3.61741424802111, 5.05479452054795,
> 3.68095238095238, 1.875, 5.25, 0)
> 
> ks.test(a,b)
> 
> data:  a and b
> D = 0.58333, p-value = 0.0337
> alternative hypothesis: two-sided
> 
> Warning message:
> In ks.test(a, b) :
> impossible to calculate exact exact value with ex-aequos
> 
> Does the p-value is correct ? Otherwise, how to solve this issue ?
> Thanks a lot.
The warning arises, not because you have "0" values as such,
but because there are repeated values (which happen to be 0).

The K-S test is designed for continuous random variables, for
which the probability of repeated values is (theoretically) zero:
theoretically, they can't happen.
>From the help page ?ks.test :
"The presence of ties always generates a warning, since continuous
distributions do not generate them. If the ties arose from
rounding the tests may be approximately valid, but even modest
amounts of rounding can have a significant effect on the
calculated statistic."



in view of the fact that your sample 'a' has three zeros along with
nine other vauwes which are all different from 0 (and all *very*
different from 0 except for 0.292452830188679), along with the fact
that your sample 'b' has 11 values all *very* different from 0.
and pne finall value equal to 0; together also with the fact that
in each sample the '0' values occur at the end, stringly suggests
that the data source is not such that a K-D test is auitasble.

The K-S test is a non-parametric test for whether
  a) a given sample comes from na given kind of distribiution;
or
  v) two samples are drwn from the same distribition.
In either case, it is assumed that the sample values are drawn
independently of each other; if there is some reason why they
may not be independent of each other, the test os not valid.

You say "I'm trying to use ks.test in order to compare two curve".
When I ezecute
  plot(a)
  plot(b)
on your data, I see (approximately) in each case a rise from a
medium vale (~2 or ~3) to a higher vale {~6 or ~10) followed
by a decline down to an exact 0.

This is not the sort of situation that the K-S test is for!
Hoping this helps,
Ted.

Thanks a lot for this reply

'a' is a simulated data while 'b' is empirical data.
Other than correlation, how to check ressemblence between these two curve
in terms of :
Amplitude in each row 1...12
Evolution and variability from 1 to 12

Thanks !


Le lundi 10 d?cembre 2018, Ted Harding <ted.harding at wlandres.net> a
?crit :
> On Mon, 2018-12-10 at 22:17 +0100, Fatma Ell wrote:
> > Dear all,
> > I'm trying to use ks.test in order to compare two curve. I've
0 values i
> > think this is why I have the follonwing warnings :impossible to
calculate
> > exact exact value with  ex-aequos
> >
> > a=c(3.02040816326531, 7.95918367346939, 10.6162790697674,
> 4.64150943396226,
> > 1.86538461538462, 1.125, 1.01020408163265, 1.2093023255814,
> > 0.292452830188679,
> > 0, 0, 0)
> > b=c(2.30769230769231, 4.19252873563218, 5.81924882629108,
> 6.2248243559719,
> > 5.02682926829268, 4.50728862973761, 3.61741424802111,
5.05479452054795,
> > 3.68095238095238, 1.875, 5.25, 0)
> >
> > ks.test(a,b)
> >
> > data:  a and b
> > D = 0.58333, p-value = 0.0337
> > alternative hypothesis: two-sided
> >
> > Warning message:
> > In ks.test(a, b) :
> > impossible to calculate exact exact value with ex-aequos
> >
> > Does the p-value is correct ? Otherwise, how to solve this issue ?
> > Thanks a lot.
>
> The warning arises, not because you have "0" values as such,
> but because there are repeated values (which happen to be 0).
>
> The K-S test is designed for continuous random variables, for
> which the probability of repeated values is (theoretically) zero:
> theoretically, they can't happen.
>
> >From the help page ?ks.test :
>
> "The presence of ties always generates a warning, since continuous
> distributions do not generate them. If the ties arose from
> rounding the tests may be approximately valid, but even modest
> amounts of rounding can have a significant effect on the
> calculated statistic."
>
>
>
> in view of the fact that your sample 'a' has three zeros along with
> nine other vauwes which are all different from 0 (and all *very*
> different from 0 except for 0.292452830188679), along with the fact
> that your sample 'b' has 11 values all *very* different from 0.
> and pne finall value equal to 0; together also with the fact that
> in each sample the '0' values occur at the end, stringly suggests
> that the data source is not such that a K-D test is auitasble.
>
> The K-S test is a non-parametric test for whether
>   a) a given sample comes from na given kind of distribiution;
> or
>   v) two samples are drwn from the same distribition.
> In either case, it is assumed that the sample values are drawn
> independently of each other; if there is some reason why they
> may not be independent of each other, the test os not valid.
>
> You say "I'm trying to use ks.test in order to compare two
curve".
> When I ezecute
>   plot(a)
>   plot(b)
> on your data, I see (approximately) in each case a rise from a
> medium vale (~2 or ~3) to a higher vale {~6 or ~10) followed
> by a decline down to an exact 0.
>
> This is not the sort of situation that the K-S test is for!
> Hoping this helps,
> Ted.
>
>
	[[alternative HTML version deleted]]