Re: [Jprogramming] density plot tick marks misaligned

2022-11-04 Thread Raul Miller
Ah, thank you. It seems so obvious now.

Anyways, I would insert the change I mentioned in
http://jsoftware.com/pipermail/programming/2022-November/061546.html
after line 2013, here:
https://github.com/jsoftware/graphics_plot/blob/master/jzplot.ijs#L2013

Thanks,

-- 
Raul

On Fri, Nov 4, 2022 at 8:43 PM chris burke  wrote:
>
> > That said, currently I do not know which github repository contains 
> > jzplot.ijs.
>
> https://github.com/jsoftware/graphics_plot
>
> On Fri, Nov 4, 2022 at 4:32 PM Raul Miller  wrote:
> >
> > Well, that's certainly doable. But, a few hours ago, I did not know
> > enough about how plot was implemented to suggest an approach there.
> >
> > That said, I think that adding:
> >
> >  if. (<'density') e. types do.
> >   'xmin xmax ymin ymax'=. (xmin,xmax,ymin,ymax)+4$0.5*_1 1
> > end.
> >
> > just before the final line of getdatarange in jzplot would do the trick.
> >
> > (That said, currently I do not know which github repository contains
> > jzplot.ijs.)
> >
> > --
> > Raul
> >
> > On Fri, Nov 4, 2022 at 4:44 PM Jan-Pieter Jacobs
> >  wrote:
> > >
> > > Well, the axes are useful, the same way they are useful for e.g. a bar
> > > plot: one would like to see which fields are in which rows and columns
> > > (without having to count them).
> > > My use case is showing a similarity matrix, in which I want to see which
> > > item in the rows corresponds with which one in the columns.
> > >
> > > I don't think the axes make sense at all (certainly not the non-integer
> > > intermediates like 2.5), they should just list the cell indices, centred 
> > > on
> > > the cells.
> > >
> > > I can't personally see a reason why it wouldn't be better to centre the
> > > ticks on the cells.
> > >
> > > Jan-Pieter
> > >
> > > On Fri, 4 Nov 2022, 14:57 Raul Miller,  wrote:
> > >
> > > > Hmm... I think you're asking for the x and y axis labels for that plot
> > > > to range from  0 to 5. This would result in blocks centered over the
> > > > integers corresponding to the numbered values.
> > > >
> > > > But the axes don't carry much (if any) information here. So maybe it
> > > > would be better to use
> > > >
> > > >'density;noaxes' plot i.5 5
> > > >
> > > > ?
> > > >
> > > > --
> > > > Raul
> > > >
> > > > On Fri, Nov 4, 2022 at 4:36 AM Jan-Pieter Jacobs
> > > >  wrote:
> > > > >
> > > > > Hi all,
> > > > >
> > > > > I noticed the tick marks in the plot addon's density plot are 
> > > > > misaligned,
> > > > > e.g.
> > > > >
> > > > > load 'plot'
> > > > > 'density' plot i. 5 5
> > > > >
> > > > > shows 5 coloured cells, and axes range from 0 to 4. But the 
> > > > > corresponding
> > > > > tick marks are aligned to the left of the cell in column 0, and the 
> > > > > right
> > > > > of the cell in column 4.
> > > > >
> > > > > I would find it more clear to have the ticks centred on the cells for 
> > > > > all
> > > > > rows and columns.
> > > > >
> > > > > What do you think? Or am I missing some useful case of the current
> > > > > behaviour?
> > > > >
> > > > > Thanks,
> > > > >
> > > > > Jan-Pieter
> > > > > --
> > > > > For information about J forums see http://www.jsoftware.com/forums.htm
> > > > --
> > > > For information about J forums see http://www.jsoftware.com/forums.htm
> > > >
> > > --
> > > For information about J forums see http://www.jsoftware.com/forums.htm
> > --
> > For information about J forums see http://www.jsoftware.com/forums.htm
> --
> For information about J forums see http://www.jsoftware.com/forums.htm
--
For information about J forums see http://www.jsoftware.com/forums.htm


Re: [Jprogramming] density plot tick marks misaligned

2022-11-04 Thread chris burke
> That said, currently I do not know which github repository contains 
> jzplot.ijs.

https://github.com/jsoftware/graphics_plot

On Fri, Nov 4, 2022 at 4:32 PM Raul Miller  wrote:
>
> Well, that's certainly doable. But, a few hours ago, I did not know
> enough about how plot was implemented to suggest an approach there.
>
> That said, I think that adding:
>
>  if. (<'density') e. types do.
>   'xmin xmax ymin ymax'=. (xmin,xmax,ymin,ymax)+4$0.5*_1 1
> end.
>
> just before the final line of getdatarange in jzplot would do the trick.
>
> (That said, currently I do not know which github repository contains
> jzplot.ijs.)
>
> --
> Raul
>
> On Fri, Nov 4, 2022 at 4:44 PM Jan-Pieter Jacobs
>  wrote:
> >
> > Well, the axes are useful, the same way they are useful for e.g. a bar
> > plot: one would like to see which fields are in which rows and columns
> > (without having to count them).
> > My use case is showing a similarity matrix, in which I want to see which
> > item in the rows corresponds with which one in the columns.
> >
> > I don't think the axes make sense at all (certainly not the non-integer
> > intermediates like 2.5), they should just list the cell indices, centred on
> > the cells.
> >
> > I can't personally see a reason why it wouldn't be better to centre the
> > ticks on the cells.
> >
> > Jan-Pieter
> >
> > On Fri, 4 Nov 2022, 14:57 Raul Miller,  wrote:
> >
> > > Hmm... I think you're asking for the x and y axis labels for that plot
> > > to range from  0 to 5. This would result in blocks centered over the
> > > integers corresponding to the numbered values.
> > >
> > > But the axes don't carry much (if any) information here. So maybe it
> > > would be better to use
> > >
> > >'density;noaxes' plot i.5 5
> > >
> > > ?
> > >
> > > --
> > > Raul
> > >
> > > On Fri, Nov 4, 2022 at 4:36 AM Jan-Pieter Jacobs
> > >  wrote:
> > > >
> > > > Hi all,
> > > >
> > > > I noticed the tick marks in the plot addon's density plot are 
> > > > misaligned,
> > > > e.g.
> > > >
> > > > load 'plot'
> > > > 'density' plot i. 5 5
> > > >
> > > > shows 5 coloured cells, and axes range from 0 to 4. But the 
> > > > corresponding
> > > > tick marks are aligned to the left of the cell in column 0, and the 
> > > > right
> > > > of the cell in column 4.
> > > >
> > > > I would find it more clear to have the ticks centred on the cells for 
> > > > all
> > > > rows and columns.
> > > >
> > > > What do you think? Or am I missing some useful case of the current
> > > > behaviour?
> > > >
> > > > Thanks,
> > > >
> > > > Jan-Pieter
> > > > --
> > > > For information about J forums see http://www.jsoftware.com/forums.htm
> > > --
> > > For information about J forums see http://www.jsoftware.com/forums.htm
> > >
> > --
> > For information about J forums see http://www.jsoftware.com/forums.htm
> --
> For information about J forums see http://www.jsoftware.com/forums.htm
--
For information about J forums see http://www.jsoftware.com/forums.htm


Re: [Jprogramming] density plot tick marks misaligned

2022-11-04 Thread Raul Miller
Well, that's certainly doable. But, a few hours ago, I did not know
enough about how plot was implemented to suggest an approach there.

That said, I think that adding:

 if. (<'density') e. types do.
  'xmin xmax ymin ymax'=. (xmin,xmax,ymin,ymax)+4$0.5*_1 1
end.

just before the final line of getdatarange in jzplot would do the trick.

(That said, currently I do not know which github repository contains
jzplot.ijs.)

-- 
Raul

On Fri, Nov 4, 2022 at 4:44 PM Jan-Pieter Jacobs
 wrote:
>
> Well, the axes are useful, the same way they are useful for e.g. a bar
> plot: one would like to see which fields are in which rows and columns
> (without having to count them).
> My use case is showing a similarity matrix, in which I want to see which
> item in the rows corresponds with which one in the columns.
>
> I don't think the axes make sense at all (certainly not the non-integer
> intermediates like 2.5), they should just list the cell indices, centred on
> the cells.
>
> I can't personally see a reason why it wouldn't be better to centre the
> ticks on the cells.
>
> Jan-Pieter
>
> On Fri, 4 Nov 2022, 14:57 Raul Miller,  wrote:
>
> > Hmm... I think you're asking for the x and y axis labels for that plot
> > to range from  0 to 5. This would result in blocks centered over the
> > integers corresponding to the numbered values.
> >
> > But the axes don't carry much (if any) information here. So maybe it
> > would be better to use
> >
> >'density;noaxes' plot i.5 5
> >
> > ?
> >
> > --
> > Raul
> >
> > On Fri, Nov 4, 2022 at 4:36 AM Jan-Pieter Jacobs
> >  wrote:
> > >
> > > Hi all,
> > >
> > > I noticed the tick marks in the plot addon's density plot are misaligned,
> > > e.g.
> > >
> > > load 'plot'
> > > 'density' plot i. 5 5
> > >
> > > shows 5 coloured cells, and axes range from 0 to 4. But the corresponding
> > > tick marks are aligned to the left of the cell in column 0, and the right
> > > of the cell in column 4.
> > >
> > > I would find it more clear to have the ticks centred on the cells for all
> > > rows and columns.
> > >
> > > What do you think? Or am I missing some useful case of the current
> > > behaviour?
> > >
> > > Thanks,
> > >
> > > Jan-Pieter
> > > --
> > > For information about J forums see http://www.jsoftware.com/forums.htm
> > --
> > For information about J forums see http://www.jsoftware.com/forums.htm
> >
> --
> For information about J forums see http://www.jsoftware.com/forums.htm
--
For information about J forums see http://www.jsoftware.com/forums.htm


Re: [Jprogramming] density plot tick marks misaligned

2022-11-04 Thread Jan-Pieter Jacobs
Well, the axes are useful, the same way they are useful for e.g. a bar
plot: one would like to see which fields are in which rows and columns
(without having to count them).
My use case is showing a similarity matrix, in which I want to see which
item in the rows corresponds with which one in the columns.

I don't think the axes make sense at all (certainly not the non-integer
intermediates like 2.5), they should just list the cell indices, centred on
the cells.

I can't personally see a reason why it wouldn't be better to centre the
ticks on the cells.

Jan-Pieter

On Fri, 4 Nov 2022, 14:57 Raul Miller,  wrote:

> Hmm... I think you're asking for the x and y axis labels for that plot
> to range from  0 to 5. This would result in blocks centered over the
> integers corresponding to the numbered values.
>
> But the axes don't carry much (if any) information here. So maybe it
> would be better to use
>
>'density;noaxes' plot i.5 5
>
> ?
>
> --
> Raul
>
> On Fri, Nov 4, 2022 at 4:36 AM Jan-Pieter Jacobs
>  wrote:
> >
> > Hi all,
> >
> > I noticed the tick marks in the plot addon's density plot are misaligned,
> > e.g.
> >
> > load 'plot'
> > 'density' plot i. 5 5
> >
> > shows 5 coloured cells, and axes range from 0 to 4. But the corresponding
> > tick marks are aligned to the left of the cell in column 0, and the right
> > of the cell in column 4.
> >
> > I would find it more clear to have the ticks centred on the cells for all
> > rows and columns.
> >
> > What do you think? Or am I missing some useful case of the current
> > behaviour?
> >
> > Thanks,
> >
> > Jan-Pieter
> > --
> > For information about J forums see http://www.jsoftware.com/forums.htm
> --
> For information about J forums see http://www.jsoftware.com/forums.htm
>
--
For information about J forums see http://www.jsoftware.com/forums.htm


Re: [Jprogramming] density plot tick marks misaligned

2022-11-04 Thread Raul Miller
Hmm... I think you're asking for the x and y axis labels for that plot
to range from  0 to 5. This would result in blocks centered over the
integers corresponding to the numbered values.

But the axes don't carry much (if any) information here. So maybe it
would be better to use

   'density;noaxes' plot i.5 5

?

-- 
Raul

On Fri, Nov 4, 2022 at 4:36 AM Jan-Pieter Jacobs
 wrote:
>
> Hi all,
>
> I noticed the tick marks in the plot addon's density plot are misaligned,
> e.g.
>
> load 'plot'
> 'density' plot i. 5 5
>
> shows 5 coloured cells, and axes range from 0 to 4. But the corresponding
> tick marks are aligned to the left of the cell in column 0, and the right
> of the cell in column 4.
>
> I would find it more clear to have the ticks centred on the cells for all
> rows and columns.
>
> What do you think? Or am I missing some useful case of the current
> behaviour?
>
> Thanks,
>
> Jan-Pieter
> --
> For information about J forums see http://www.jsoftware.com/forums.htm
--
For information about J forums see http://www.jsoftware.com/forums.htm


Re: [Jprogramming] density plot tick marks misaligned

2022-11-04 Thread David Lambert
The ticks look correct, placing 5 panels into 4 zones.  Perhaps the 
scales should run from _0.5 to 4.5,


Well this doesn't quite work:

'density;yrange _0.5, 4.5' plot i. 5 5

(Engine: j904/j64avx2/linux  Beta-e)


ate: Fri, 4 Nov 2022 09:35:53 +0100
From: Jan-Pieter Jacobs
To:programm...@jsoftware.com
Subject: [Jprogramming] density plot tick marks misaligned
Message-ID:

Content-Type: text/plain; charset="UTF-8"

Hi all,

I noticed the tick marks in the plot addon's density plot are misaligned,
e.g.

load 'plot'
'density' plot i. 5 5

shows 5 coloured cells, and axes range from 0 to 4. But the corresponding
tick marks are aligned to the left of the cell in column 0, and the right
of the cell in column 4.

I would find it more clear to have the ticks centred on the cells for all
rows and columns.

What do you think? Or am I missing some useful case of the current
behaviour?

Thanks,

Jan-Pieter


--
For information about J forums see http://www.jsoftware.com/forums.htm


[Jprogramming] density plot tick marks misaligned

2022-11-04 Thread Jan-Pieter Jacobs
Hi all,

I noticed the tick marks in the plot addon's density plot are misaligned,
e.g.

load 'plot'
'density' plot i. 5 5

shows 5 coloured cells, and axes range from 0 to 4. But the corresponding
tick marks are aligned to the left of the cell in column 0, and the right
of the cell in column 4.

I would find it more clear to have the ticks centred on the cells for all
rows and columns.

What do you think? Or am I missing some useful case of the current
behaviour?

Thanks,

Jan-Pieter
--
For information about J forums see http://www.jsoftware.com/forums.htm