[GRASS-user] r.quantile: what it exaclty do?

Hi
I tirend used r.quantile on SRTM data for Poland (min value about to 0, max value about to 2500 m a.s.l)

r.quantile input=Polska@PERMANENT quantiles=4 percentiles=0.001,0.01,0,1,0.25,0.50,0.75,0.90,0.99,0.999 bins=1000000

and I recived:

0:0.000000:0.000000
1:0.001000:0.000000
2:0.010000:0.000000
3:0.250000:0.000000
4:0.500000:1.000000
5:0.750000:2.000000
6:0.900000:3.000000
7:0.990000:3.000000
8:0.999000:3.000000
9:1.000000:3.000000

It looks like the results are divided by 1000 and rounded to the nearest integer.

Did I something wrong?

Jarek

On Fri, Jan 2, 2009 at 4:55 AM, Jarek Jasiewicz <jarekj@amu.edu.pl> wrote:

Hi
I tirend used r.quantile on SRTM data for Poland (min value about to 0, max
value about to 2500 m a.s.l)

r.quantile input=Polska@PERMANENT quantiles=4
percentiles=0.001,0.01,0,1,0.25,0.50,0.75,0.90,0.99,0.999 bins=1000000

and I recived:

0:0.000000:0.000000
1:0.001000:0.000000
2:0.010000:0.000000
3:0.250000:0.000000
4:0.500000:1.000000
5:0.750000:2.000000
6:0.900000:3.000000
7:0.990000:3.000000
8:0.999000:3.000000
9:1.000000:3.000000

It looks like the results are divided by 1000 and rounded to the nearest
integer.

Did I something wrong?

Jarek

I thought that the user supplies one of [quantiles=] | [percentiles=]
... could that be related to the odd output. I have verified (with R)
that r.quantile can compute correct quantiles... however, on my system
I need to set bins=100 or so, as the higher i set 'bins' odd things
would happen.

Glynn should know for sure.

Dylan

Dylan Beaudette pisze:

On Fri, Jan 2, 2009 at 4:55 AM, Jarek Jasiewicz <jarekj@amu.edu.pl> wrote:
  

Hi
I tirend used r.quantile on SRTM data for Poland (min value about to 0, max
value about to 2500 m a.s.l)

r.quantile input=Polska@PERMANENT quantiles=4
percentiles=0.001,0.01,0,1,0.25,0.50,0.75,0.90,0.99,0.999 bins=1000000

and I recived:

0:0.000000:0.000000
1:0.001000:0.000000
2:0.010000:0.000000
3:0.250000:0.000000
4:0.500000:1.000000
5:0.750000:2.000000
6:0.900000:3.000000
7:0.990000:3.000000
8:0.999000:3.000000
9:1.000000:3.000000

It looks like the results are divided by 1000 and rounded to the nearest
integer.

Did I something wrong?

Jarek

I thought that the user supplies one of [quantiles=] | [percentiles=]
... could that be related to the odd output. I have verified (with R)
that r.quantile can compute correct quantiles... however, on my system
I need to set bins=100 or so, as the higher i set 'bins' odd things
would happen.

Glynn should know for sure.

Dylan
  

Thanks
it really seems that the problem was in default number of bins

On Fri, Jan 2, 2009 at 8:40 PM, Dylan Beaudette
<dylan.beaudette@gmail.com> wrote:

On Fri, Jan 2, 2009 at 4:55 AM, Jarek Jasiewicz <jarekj@amu.edu.pl> wrote:

Hi
I tirend used r.quantile on SRTM data for Poland (min value about to 0, max
value about to 2500 m a.s.l)

r.quantile input=Polska@PERMANENT quantiles=4
percentiles=0.001,0.01,0,1,0.25,0.50,0.75,0.90,0.99,0.999 bins=1000000

and I recived:

0:0.000000:0.000000
1:0.001000:0.000000
2:0.010000:0.000000
3:0.250000:0.000000
4:0.500000:1.000000
5:0.750000:2.000000
6:0.900000:3.000000
7:0.990000:3.000000
8:0.999000:3.000000
9:1.000000:3.000000

It looks like the results are divided by 1000 and rounded to the nearest
integer.

Did I something wrong?

Jarek

I thought that the user supplies one of [quantiles=] | [percentiles=]
... could that be related to the odd output. I have verified (with R)
that r.quantile can compute correct quantiles... however, on my system
I need to set bins=100 or so, as the higher i set 'bins' odd things
would happen.

Glynn should know for sure.

Can you please file a bug report on this? Preferable with a
Spearfish/NC reproducible example?

Thanks
Markus