[R] Problems with ks.test()

[Jochen1980]

Hi, 

I got two data point vectors. Now I want to make a ks.test(). I you print 
both vectors you will see, that they fit pretty fine. Here is a picture:
http://www.jochen-bauer.net/downloads/kstest-r-help-list-plot.png

As you can see there is one histogram and moreover there is the gumbel
density 
function plotted. Now I took to bin-mids and the bin-height for vector1 and 
computed the distribution-values to all bin-mids as vector2. 

I pass these two vectors to ks.test(). Are those the right vectors, if I
want
to decide afterwards, if my experiment-data is gumbel-distributed? 

Surprisingly the p-value changes tremendously if I calculate more digits out
of 
my theoretical formula. If I round to 0 digits, p is 1, if I round to 4
digits,
p drops to 0 - how could this happen, I thought more digits will bring more
accurate results?!

XXXX Case 0 digits: XXXXXXXXXXXXXXXXXXXXXXXXXXX 
  [1]   0   0   0   0   0  24  74  98 133 147 134 120  89  69  46  31  16  
7
 [19]   7   3   2   0   0   0   0   0   0   0   0   0   0   0   0   0   0  
0
 [37]   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0  
0
 [55]   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0  
0
 [73]   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0  
0
 [91]   0   0   0   0   0   0   0   0   0   0
  [1]   0   0   0   0   1  10  49 113 160 168 147 113  81  55  37  24  15 
10
 [19]   6   4   2   2   1   1   0   0   0   0   0   0   0   0   0   0   0  
0
 [37]   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0  
0
 [55]   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0  
0
 [73]   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0  
0
 [91]   0   0   0   0   0   0   0   0   0   0
[1] "Ergebnisse"
[1] "Analyse der Eingangsdaten"
[1] "Mean:  0.104537195"
[1] "SAbw.:  0.0277657985898433"
[1] "Parameter-Berechnung der Daten bei angenommener Gumbelverteilung"
[1] "Mue:  0.0920411082987717"
[1] "Beta:  0.0216489043196013"
[1] "KS-Test ->  1000  Werte,  100  Bins, x: Klassenmitten, y1, y2 =
Histogrammhöhen"
[1] "KST D:  0.04"
[1] "KST P:  1"

XXX Case 4 digits: XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
  [1]   0   0   0   0   0  24  74  98 133 147 134 120  89  69  46  31  16  
7
 [19]   7   3   2   0   0   0   0   0   0   0   0   0   0   0   0   0   0  
0
 [37]   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0  
0
 [55]   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0  
0
 [73]   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0  
0
 [91]   0   0   0   0   0   0   0   0   0   0
  [1]   0.000   0.000   0.000   0.006   0.622  10.094  49.271 112.776
160.174
 [10] 168.419 146.527 113.137  81.026  55.344  36.690  23.870  15.347  
9.793
 [19]   6.220   3.939   2.490   1.572   0.992   0.625   0.394   0.248  
0.157
 [28]   0.099   0.062   0.039   0.025   0.016   0.010   0.006   0.004  
0.002
 [37]   0.002   0.001   0.001   0.000   0.000   0.000   0.000   0.000  
0.000
 [46]   0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.000  
0.000
 [55]   0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.000  
0.000
 [64]   0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.000  
0.000
 [73]   0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.000  
0.000
 [82]   0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.000  
0.000
 [91]   0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.000  
0.000
[100]   0.000
[1] "Ergebnisse"
[1] "Analyse der Eingangsdaten"
[1] "Mean:  0.104537195"
[1] "SAbw.:  0.0277657985898433"
[1] "Parameter-Berechnung der Daten bei angenommener Gumbelverteilung"
[1] "Mue:  0.0920411082987717"
[1] "Beta:  0.0216489043196013"
[1] "KS-Test ->  1000  Werte,  100  Bins, x: Klassenmitten, y1, y2 =
Histogrammhöhen"
[1] "KST D:  0.2"
[1] "KST P:  0.0366"

Thanks in advance for some help.
Jochen

--
View this message in context: http://r.789695.n4.nabble.com/Problems-with-ks-test-tp3703469p3703469.html
Sent from the R help mailing list archive at Nabble.com.