Extra y axis is not rendered when LogAxis is used

I am trying to put up an extra y axis with log scale using the following code:

p = figure(plot_width=800, plot_height=400, toolbar_location=None)

p.line([1,2,3], [1,2,3])
p.y_range = Range1d(0,4)

p.circle([1,2,3],[10,100,1000], y_range_name=‘Y’)
p.extra_y_ranges[‘Y’]=Range1d(0,1000)
p.add_layout(LogAxis(y_range_name=‘Y’, axis_label=‘H’), ‘right’)

show§

``

When I use LinearAxis, the y axis is rendered well along with ticks and lables. However with LogAxis, only axis label is shown.

What goes wrong in the above code?

Hi,

···

On Wed, Aug 30, 2017 at 9:53 AM, Sijun Cho [email protected] wrote:

I am trying to put up an extra y axis with log scale using the following code:

p = figure(plot_width=800, plot_height=400, toolbar_location=None)

p.line([1,2,3], [1,2,3])
p.y_range = Range1d(0,4)

p.circle([1,2,3],[10,100,1000], y_range_name=‘Y’)
p.extra_y_ranges[‘Y’]=Range1d(0,1000)
p.add_layout(LogAxis(y_range_name=‘Y’, axis_label=‘H’), ‘right’)

show§

``

When I use LinearAxis, the y axis is rendered well along with ticks and lables. However with LogAxis, only axis label is shown.

What goes wrong in the above code?

the problem is in lack of log scale on the y-axis for circle glyph. A scale is what’s actually responsible for mapping data to screen coordinates. It looks like currently it’s not possible to configure a custom scale along a custom range. I’m not sure right now if there is any inherent problem with such setup, or this is just an oversight.

Mateusz

You received this message because you are subscribed to the Google Groups “Bokeh Discussion - Public” group.

To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].

To post to this group, send email to [email protected].

To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/46adfb00-2a9f-41f5-a33f-cebb71afb974%40continuum.io.

For more options, visit https://groups.google.com/a/continuum.io/d/optout.

Thank you Mateusz for replying.

If this is the limitaion at the moment, perhaps I need to make a new feature request at the Github.

Sijun

2017년 8월 30일 수요일 오후 11시 13분 35초 UTC+9, mateusz.paprocki 님의 말:

···

Hi,

On Wed, Aug 30, 2017 at 9:53 AM, Sijun Cho [email protected] wrote:

I am trying to put up an extra y axis with log scale using the following code:

p = figure(plot_width=800, plot_height=400, toolbar_location=None)

p.line([1,2,3], [1,2,3])
p.y_range = Range1d(0,4)

p.circle([1,2,3],[10,100,1000], y_range_name=‘Y’)
p.extra_y_ranges[‘Y’]=Range1d(0,1000)
p.add_layout(LogAxis(y_range_name=‘Y’, axis_label=‘H’), ‘right’)

show§

``

When I use LinearAxis, the y axis is rendered well along with ticks and lables. However with LogAxis, only axis label is shown.

What goes wrong in the above code?

the problem is in lack of log scale on the y-axis for circle glyph. A scale is what’s actually responsible for mapping data to screen coordinates. It looks like currently it’s not possible to configure a custom scale along a custom range. I’m not sure right now if there is any inherent problem with such setup, or this is just an oversight.

Mateusz

You received this message because you are subscribed to the Google Groups “Bokeh Discussion - Public” group.

To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].

To post to this group, send email to [email protected].

To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/46adfb00-2a9f-41f5-a33f-cebb71afb974%40continuum.io.

For more options, visit https://groups.google.com/a/continuum.io/d/optout.

What is the use case for combining a linear and log axis simultaneously? How should they behave together under panning actions?

Thanks,

Bryan

···

On Aug 31, 2017, at 02:43, Sijun Cho <[email protected]> wrote:

Thank you Mateusz for replying.

If this is the limitaion at the moment, perhaps I need to make a new feature request at the Github.

Sijun

2017년 8월 30일 수요일 오후 11시 13분 35초 UTC+9, mateusz.paprocki 님의 말:
Hi,

On Wed, Aug 30, 2017 at 9:53 AM, Sijun Cho <[email protected]> wrote:
I am trying to put up an extra y axis with log scale using the following code:

p = figure(plot_width=800, plot_height=400, toolbar_location=None)

p.line([1,2,3], [1,2,3])
p.y_range = Range1d(0,4)

p.circle([1,2,3],[10,100,1000], y_range_name='Y')
p.extra_y_ranges['Y']=Range1d(0,1000)
p.add_layout(LogAxis(y_range_name='Y', axis_label='H'), 'right')

show(p)

When I use LinearAxis, the y axis is rendered well along with ticks and lables. However with LogAxis, only axis label is shown.
What goes wrong in the above code?

the problem is in lack of log scale on the y-axis for circle glyph. A scale is what's actually responsible for mapping data to screen coordinates. It looks like currently it's not possible to configure a custom scale along a custom range. I'm not sure right now if there is any inherent problem with such setup, or this is just an oversight.

Mateusz

--
You received this message because you are subscribed to the Google Groups "Bokeh Discussion - Public" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
To post to this group, send email to [email protected].
To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/46adfb00-2a9f-41f5-a33f-cebb71afb974%40continuum.io.
For more options, visit https://groups.google.com/a/continuum.io/d/optout.

--
You received this message because you are subscribed to the Google Groups "Bokeh Discussion - Public" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
To post to this group, send email to [email protected].
To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/7d60fab6-dd6b-4ac4-a125-245137a95850%40continuum.io.
For more options, visit https://groups.google.com/a/continuum.io/d/optout.

Thank you Bryan for replying.

I obtained the below graph using matplotlib.

The x-axis is categorical and each tick corresponds to company ID of Wi-Fi Access Points. (not shown because the count is very large)

The left y-axis is for red curve (actually dots) and indicates degree of eccentricity that each AP device shows.

However, the ‘eccentricity’ is not the only consideration. If a group of APs of a particular ID produces relatively small number of error messages, then it will count for only that much portion of customer complaints received, i.e., I wanted to focus on the area in the graph with relatively large error messages, since it is the area that has greater influence on overall market satisfaction.

The green dots in the graph indicates the number of error messages and initially I tried to draw them in linear scale but since the difference of the maximum and minimun values was so large, when rendered, it most of the points were near the floor of the plot.

So I wanted to warp the scale using logarithm. This is maybe a very corner use case.

I never thought about how the axes behave under spanning action! What if adding an option for disabling of spanning action (making the plot static) and thereby allow for linear + logscale together like the example plot below?

Thanks,

Sijun

2017년 8월 31일 목요일 오후 11시 37분 27초 UTC+9, Bryan Van de ven 님의 말:

···

What is the use case for combining a linear and log axis simultaneously? How should they behave together under panning actions?

Thanks,

Bryan

On Aug 31, 2017, at 02:43, Sijun Cho [email protected] wrote:

Thank you Mateusz for replying.

If this is the limitaion at the moment, perhaps I need to make a new feature request at the Github.

Sijun

2017년 8월 30일 수요일 오후 11시 13분 35초 UTC+9, mateusz.paprocki 님의 말:

Hi,

On Wed, Aug 30, 2017 at 9:53 AM, Sijun Cho [email protected] wrote:

I am trying to put up an extra y axis with log scale using the following code:

p = figure(plot_width=800, plot_height=400, toolbar_location=None)

p.line([1,2,3], [1,2,3])

p.y_range = Range1d(0,4)

p.circle([1,2,3],[10,100,1000], y_range_name=‘Y’)

p.extra_y_ranges[‘Y’]=Range1d(0,1000)

p.add_layout(LogAxis(y_range_name=‘Y’, axis_label=‘H’), ‘right’)

show§

When I use LinearAxis, the y axis is rendered well along with ticks and lables. However with LogAxis, only axis label is shown.

What goes wrong in the above code?

the problem is in lack of log scale on the y-axis for circle glyph. A scale is what’s actually responsible for mapping data to screen coordinates. It looks like currently it’s not possible to configure a custom scale along a custom range. I’m not sure right now if there is any inherent problem with such setup, or this is just an oversight.

Mateusz


You received this message because you are subscribed to the Google Groups “Bokeh Discussion - Public” group.

To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].

To post to this group, send email to [email protected].

To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/46adfb00-2a9f-41f5-a33f-cebb71afb974%40continuum.io.

For more options, visit https://groups.google.com/a/continuum.io/d/optout.


You received this message because you are subscribed to the Google Groups “Bokeh Discussion - Public” group.

To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].

To post to this group, send email to [email protected].

To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/7d60fab6-dd6b-4ac4-a125-245137a95850%40continuum.io.

For more options, visit https://groups.google.com/a/continuum.io/d/optout.

Hi,

As a workaround, instead of plotting #errors on a log scale, what about plotting log(#errors) on a linear scale?

Thanks,

Bryan

···

On Sep 1, 2017, at 02:45, Sijun Cho <[email protected]> wrote:

Thank you Bryan for replying.

I obtained the below graph using matplotlib.

The x-axis is categorical and each tick corresponds to company ID of Wi-Fi Access Points. (not shown because the count is very large)

The left y-axis is for red curve (actually dots) and indicates degree of eccentricity that each AP device shows.

However, the 'eccentricity' is not the only consideration. If a group of APs of a particular ID produces relatively small number of error messages, then it will count for only that much portion of customer complaints received, i.e., I wanted to focus on the area in the graph with relatively large error messages, since it is the area that has greater influence on overall market satisfaction.

The green dots in the graph indicates the number of error messages and initially I tried to draw them in linear scale but since the difference of the maximum and minimun values was so large, when rendered, it most of the points were near the floor of the plot.

So I wanted to warp the scale using logarithm. This is maybe a very corner use case.

I never thought about how the axes behave under spanning action! What if adding an option for disabling of spanning action (making the plot static) and thereby allow for linear + logscale together like the example plot below?

Thanks,

Sijun

2017년 8월 31일 목요일 오후 11시 37분 27초 UTC+9, Bryan Van de ven 님의 말:
What is the use case for combining a linear and log axis simultaneously? How should they behave together under panning actions?

Thanks,

Bryan

> On Aug 31, 2017, at 02:43, Sijun Cho <[email protected]> wrote:
>
> Thank you Mateusz for replying.
>
> If this is the limitaion at the moment, perhaps I need to make a new feature request at the Github.
>
> Sijun
>
> 2017년 8월 30일 수요일 오후 11시 13분 35초 UTC+9, mateusz.paprocki 님의 말:
> Hi,
>
> On Wed, Aug 30, 2017 at 9:53 AM, Sijun Cho <[email protected]> wrote:
> I am trying to put up an extra y axis with log scale using the following code:
>
> p = figure(plot_width=800, plot_height=400, toolbar_location=None)
>
>
> p.line([1,2,3], [1,2,3])
> p.y_range = Range1d(0,4)
>
>
> p.circle([1,2,3],[10,100,1000], y_range_name='Y')
> p.extra_y_ranges['Y']=Range1d(0,1000)
> p.add_layout(LogAxis(y_range_name='Y', axis_label='H'), 'right')
>
>
> show(p)
>
>
> When I use LinearAxis, the y axis is rendered well along with ticks and lables. However with LogAxis, only axis label is shown.
> What goes wrong in the above code?
>
> the problem is in lack of log scale on the y-axis for circle glyph. A scale is what's actually responsible for mapping data to screen coordinates. It looks like currently it's not possible to configure a custom scale along a custom range. I'm not sure right now if there is any inherent problem with such setup, or this is just an oversight.
>
> Mateusz
>
>
>
> --
> You received this message because you are subscribed to the Google Groups "Bokeh Discussion - Public" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
> To post to this group, send email to [email protected].
> To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/46adfb00-2a9f-41f5-a33f-cebb71afb974%40continuum.io.
> For more options, visit https://groups.google.com/a/continuum.io/d/optout.
>
>
> --
> You received this message because you are subscribed to the Google Groups "Bokeh Discussion - Public" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
> To post to this group, send email to [email protected].
> To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/7d60fab6-dd6b-4ac4-a125-245137a95850%40continuum.io.
> For more options, visit https://groups.google.com/a/continuum.io/d/optout.

--
You received this message because you are subscribed to the Google Groups "Bokeh Discussion - Public" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
To post to this group, send email to [email protected].
To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/644f2bbe-0da2-470e-870a-2bbf874403f8%40continuum.io.
For more options, visit https://groups.google.com/a/continuum.io/d/optout.

It is a great idea! Although the axis ticks locations will look a bit different from typical log scale ticks (since minor ticks are not equally spaced in a usual log scale axis), it is good enough for the purpose of merely stretching out part of the graph

Thanks,

Sijun

2017년 9월 1일 금요일 오후 10시 37분 3초 UTC+9, Bryan Van de ven 님의 말:

···

Hi,

As a workaround, instead of plotting #errors on a log scale, what about plotting log(#errors) on a linear scale?

Thanks,

Bryan

On Sep 1, 2017, at 02:45, Sijun Cho [email protected] wrote:

Thank you Bryan for replying.

I obtained the below graph using matplotlib.

The x-axis is categorical and each tick corresponds to company ID of Wi-Fi Access Points. (not shown because the count is very large)

The left y-axis is for red curve (actually dots) and indicates degree of eccentricity that each AP device shows.

However, the ‘eccentricity’ is not the only consideration. If a group of APs of a particular ID produces relatively small number of error messages, then it will count for only that much portion of customer complaints received, i.e., I wanted to focus on the area in the graph with relatively large error messages, since it is the area that has greater influence on overall market satisfaction.

The green dots in the graph indicates the number of error messages and initially I tried to draw them in linear scale but since the difference of the maximum and minimun values was so large, when rendered, it most of the points were near the floor of the plot.

So I wanted to warp the scale using logarithm. This is maybe a very corner use case.

I never thought about how the axes behave under spanning action! What if adding an option for disabling of spanning action (making the plot static) and thereby allow for linear + logscale together like the example plot below?

Thanks,

Sijun

2017년 8월 31일 목요일 오후 11시 37분 27초 UTC+9, Bryan Van de ven 님의 말:

What is the use case for combining a linear and log axis simultaneously? How should they behave together under panning actions?

Thanks,

Bryan

On Aug 31, 2017, at 02:43, Sijun Cho [email protected] wrote:

Thank you Mateusz for replying.

If this is the limitaion at the moment, perhaps I need to make a new feature request at the Github.

Sijun

2017년 8월 30일 수요일 오후 11시 13분 35초 UTC+9, mateusz.paprocki 님의 말:
Hi,

On Wed, Aug 30, 2017 at 9:53 AM, Sijun Cho [email protected] wrote:
I am trying to put up an extra y axis with log scale using the following code:

p = figure(plot_width=800, plot_height=400, toolbar_location=None)

p.line([1,2,3], [1,2,3])
p.y_range = Range1d(0,4)

p.circle([1,2,3],[10,100,1000], y_range_name=‘Y’)
p.extra_y_ranges[‘Y’]=Range1d(0,1000)
p.add_layout(LogAxis(y_range_name=‘Y’, axis_label=‘H’), ‘right’)

show§

When I use LinearAxis, the y axis is rendered well along with ticks and lables. However with LogAxis, only axis label is shown.
What goes wrong in the above code?

the problem is in lack of log scale on the y-axis for circle glyph. A scale is what’s actually responsible for mapping data to screen coordinates. It looks like currently it’s not possible to configure a custom scale along a custom range. I’m not sure right now if there is any inherent problem with such setup, or this is just an oversight.

Mateusz


You received this message because you are subscribed to the Google Groups “Bokeh Discussion - Public” group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
To post to this group, send email to [email protected].
To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/46adfb00-2a9f-41f5-a33f-cebb71afb974%40continuum.io.
For more options, visit https://groups.google.com/a/continuum.io/d/optout.


You received this message because you are subscribed to the Google Groups “Bokeh Discussion - Public” group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
To post to this group, send email to [email protected].
To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/7d60fab6-dd6b-4ac4-a125-245137a95850%40continuum.io.
For more options, visit https://groups.google.com/a/continuum.io/d/optout.


You received this message because you are subscribed to the Google Groups “Bokeh Discussion - Public” group.

To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].

To post to this group, send email to [email protected].

To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/644f2bbe-0da2-470e-870a-2bbf874403f8%40continuum.io.

For more options, visit https://groups.google.com/a/continuum.io/d/optout.

This could work but not a feasible solution. Hope somebody could figure this out soon…

···

On Friday, 1 September 2017 14:37:03 UTC+1, Bryan Van de ven wrote:

Hi,

As a workaround, instead of plotting #errors on a log scale, what about plotting log(#errors) on a linear scale?

Thanks,

Bryan

On Sep 1, 2017, at 02:45, Sijun Cho [email protected] wrote:

Thank you Bryan for replying.

I obtained the below graph using matplotlib.

The x-axis is categorical and each tick corresponds to company ID of Wi-Fi Access Points. (not shown because the count is very large)

The left y-axis is for red curve (actually dots) and indicates degree of eccentricity that each AP device shows.

However, the ‘eccentricity’ is not the only consideration. If a group of APs of a particular ID produces relatively small number of error messages, then it will count for only that much portion of customer complaints received, i.e., I wanted to focus on the area in the graph with relatively large error messages, since it is the area that has greater influence on overall market satisfaction.

The green dots in the graph indicates the number of error messages and initially I tried to draw them in linear scale but since the difference of the maximum and minimun values was so large, when rendered, it most of the points were near the floor of the plot.

So I wanted to warp the scale using logarithm. This is maybe a very corner use case.

I never thought about how the axes behave under spanning action! What if adding an option for disabling of spanning action (making the plot static) and thereby allow for linear + logscale together like the example plot below?

Thanks,

Sijun

2017년 8월 31일 목요일 오후 11시 37분 27초 UTC+9, Bryan Van de ven 님의 말:

What is the use case for combining a linear and log axis simultaneously? How should they behave together under panning actions?

Thanks,

Bryan

On Aug 31, 2017, at 02:43, Sijun Cho [email protected] wrote:

Thank you Mateusz for replying.

If this is the limitaion at the moment, perhaps I need to make a new feature request at the Github.

Sijun

2017년 8월 30일 수요일 오후 11시 13분 35초 UTC+9, mateusz.paprocki 님의 말:
Hi,

On Wed, Aug 30, 2017 at 9:53 AM, Sijun Cho [email protected] wrote:
I am trying to put up an extra y axis with log scale using the following code:

p = figure(plot_width=800, plot_height=400, toolbar_location=None)

p.line([1,2,3], [1,2,3])
p.y_range = Range1d(0,4)

p.circle([1,2,3],[10,100,1000], y_range_name=‘Y’)
p.extra_y_ranges[‘Y’]=Range1d(0,1000)
p.add_layout(LogAxis(y_range_name=‘Y’, axis_label=‘H’), ‘right’)

show§

When I use LinearAxis, the y axis is rendered well along with ticks and lables. However with LogAxis, only axis label is shown.
What goes wrong in the above code?

the problem is in lack of log scale on the y-axis for circle glyph. A scale is what’s actually responsible for mapping data to screen coordinates. It looks like currently it’s not possible to configure a custom scale along a custom range. I’m not sure right now if there is any inherent problem with such setup, or this is just an oversight.

Mateusz


You received this message because you are subscribed to the Google Groups “Bokeh Discussion - Public” group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
To post to this group, send email to [email protected].
To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/46adfb00-2a9f-41f5-a33f-cebb71afb974%40continuum.io.
For more options, visit https://groups.google.com/a/continuum.io/d/optout.


You received this message because you are subscribed to the Google Groups “Bokeh Discussion - Public” group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
To post to this group, send email to [email protected].
To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/7d60fab6-dd6b-4ac4-a125-245137a95850%40continuum.io.
For more options, visit https://groups.google.com/a/continuum.io/d/optout.


You received this message because you are subscribed to the Google Groups “Bokeh Discussion - Public” group.

To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].

To post to this group, send email to [email protected].

To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/644f2bbe-0da2-470e-870a-2bbf874403f8%40continuum.io.

For more options, visit https://groups.google.com/a/continuum.io/d/optout.

Just being realistic: there are 650+ issues and only a couple of (very overtaxed) core contributors. I doubt we will be able to prioritize this soon as it is a pretty marginal use case in my estimation, especially since the fix is unfortunately probably not trivial. Things like interactive panning are core and central to Bokeh, so disabling them in specific cases is neither trivial nor something to consider without a lot of deliberation to see how it could be done best.

Thanks,

Bryan

···

On Feb 26, 2018, at 06:20, tarik <[email protected]> wrote:

This could work but not a feasible solution. Hope somebody could figure this out soon..

On Friday, 1 September 2017 14:37:03 UTC+1, Bryan Van de ven wrote:
Hi,

As a workaround, instead of plotting #errors on a log scale, what about plotting log(#errors) on a linear scale?

Thanks,

Bryan

> On Sep 1, 2017, at 02:45, Sijun Cho <[email protected]> wrote:
>
> Thank you Bryan for replying.
>
>
> I obtained the below graph using matplotlib.
>
> The x-axis is categorical and each tick corresponds to company ID of Wi-Fi Access Points. (not shown because the count is very large)
>
> The left y-axis is for red curve (actually dots) and indicates degree of eccentricity that each AP device shows.
>
> However, the 'eccentricity' is not the only consideration. If a group of APs of a particular ID produces relatively small number of error messages, then it will count for only that much portion of customer complaints received, i.e., I wanted to focus on the area in the graph with relatively large error messages, since it is the area that has greater influence on overall market satisfaction.
>
> The green dots in the graph indicates the number of error messages and initially I tried to draw them in linear scale but since the difference of the maximum and minimun values was so large, when rendered, it most of the points were near the floor of the plot.
>
> So I wanted to warp the scale using logarithm. This is maybe a very corner use case.
>
> I never thought about how the axes behave under spanning action! What if adding an option for disabling of spanning action (making the plot static) and thereby allow for linear + logscale together like the example plot below?
>
> Thanks,
>
> Sijun
>
>
>
>
>
>
>
> 2017년 8월 31일 목요일 오후 11시 37분 27초 UTC+9, Bryan Van de ven 님의 말:
> What is the use case for combining a linear and log axis simultaneously? How should they behave together under panning actions?
>
> Thanks,
>
> Bryan
>
> > On Aug 31, 2017, at 02:43, Sijun Cho <[email protected]> wrote:
> >
> > Thank you Mateusz for replying.
> >
> > If this is the limitaion at the moment, perhaps I need to make a new feature request at the Github.
> >
> > Sijun
> >
> > 2017년 8월 30일 수요일 오후 11시 13분 35초 UTC+9, mateusz.paprocki 님의 말:
> > Hi,
> >
> > On Wed, Aug 30, 2017 at 9:53 AM, Sijun Cho <[email protected]> wrote:
> > I am trying to put up an extra y axis with log scale using the following code:
> >
> > p = figure(plot_width=800, plot_height=400, toolbar_location=None)
> >
> >
> > p.line([1,2,3], [1,2,3])
> > p.y_range = Range1d(0,4)
> >
> >
> > p.circle([1,2,3],[10,100,1000], y_range_name='Y')
> > p.extra_y_ranges['Y']=Range1d(0,1000)
> > p.add_layout(LogAxis(y_range_name='Y', axis_label='H'), 'right')
> >
> >
> > show(p)
> >
> >
> > When I use LinearAxis, the y axis is rendered well along with ticks and lables. However with LogAxis, only axis label is shown.
> > What goes wrong in the above code?
> >
> > the problem is in lack of log scale on the y-axis for circle glyph. A scale is what's actually responsible for mapping data to screen coordinates. It looks like currently it's not possible to configure a custom scale along a custom range. I'm not sure right now if there is any inherent problem with such setup, or this is just an oversight.
> >
> > Mateusz
> >
> >
> >
> > --
> > You received this message because you are subscribed to the Google Groups "Bokeh Discussion - Public" group.
> > To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
> > To post to this group, send email to [email protected].
> > To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/46adfb00-2a9f-41f5-a33f-cebb71afb974%40continuum.io.
> > For more options, visit https://groups.google.com/a/continuum.io/d/optout.
> >
> >
> > --
> > You received this message because you are subscribed to the Google Groups "Bokeh Discussion - Public" group.
> > To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
> > To post to this group, send email to [email protected].
> > To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/7d60fab6-dd6b-4ac4-a125-245137a95850%40continuum.io.
> > For more options, visit https://groups.google.com/a/continuum.io/d/optout.
>
>
> --
> You received this message because you are subscribed to the Google Groups "Bokeh Discussion - Public" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
> To post to this group, send email to [email protected].
> To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/644f2bbe-0da2-470e-870a-2bbf874403f8%40continuum.io.
> For more options, visit https://groups.google.com/a/continuum.io/d/optout.

--
You received this message because you are subscribed to the Google Groups "Bokeh Discussion - Public" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
To post to this group, send email to [email protected].
To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/0b076d0a-1eb8-47f7-b42c-6bd314cde509%40continuum.io.
For more options, visit https://groups.google.com/a/continuum.io/d/optout.

Thanks Bryan.

···

On 26 Feb 2018 19:42, “Bryan Van de ven” [email protected] wrote:

Just being realistic: there are 650+ issues and only a couple of (very overtaxed) core contributors. I doubt we will be able to prioritize this soon as it is a pretty marginal use case in my estimation, especially since the fix is unfortunately probably not trivial. Things like interactive panning are core and central to Bokeh, so disabling them in specific cases is neither trivial nor something to consider without a lot of deliberation to see how it could be done best.

Thanks,

Bryan

On Feb 26, 2018, at 06:20, tarik [email protected] wrote:

This could work but not a feasible solution. Hope somebody could figure this out soon…

On Friday, 1 September 2017 14:37:03 UTC+1, Bryan Van de ven wrote:

Hi,

As a workaround, instead of plotting #errors on a log scale, what about plotting log(#errors) on a linear scale?

Thanks,

Bryan

On Sep 1, 2017, at 02:45, Sijun Cho [email protected] wrote:

Thank you Bryan for replying.

I obtained the below graph using matplotlib.

The x-axis is categorical and each tick corresponds to company ID of Wi-Fi Access Points. (not shown because the count is very large)

The left y-axis is for red curve (actually dots) and indicates degree of eccentricity that each AP device shows.

However, the ‘eccentricity’ is not the only consideration. If a group of APs of a particular ID produces relatively small number of error messages, then it will count for only that much portion of customer complaints received, i.e., I wanted to focus on the area in the graph with relatively large error messages, since it is the area that has greater influence on overall market satisfaction.

The green dots in the graph indicates the number of error messages and initially I tried to draw them in linear scale but since the difference of the maximum and minimun values was so large, when rendered, it most of the points were near the floor of the plot.

So I wanted to warp the scale using logarithm. This is maybe a very corner use case.

I never thought about how the axes behave under spanning action! What if adding an option for disabling of spanning action (making the plot static) and thereby allow for linear + logscale together like the example plot below?

Thanks,

Sijun

2017년 8월 31일 목요일 오후 11시 37분 27초 UTC+9, Bryan Van de ven 님의 말:

What is the use case for combining a linear and log axis simultaneously? How should they behave together under panning actions?

Thanks,

Bryan

On Aug 31, 2017, at 02:43, Sijun Cho [email protected] wrote:

Thank you Mateusz for replying.

If this is the limitaion at the moment, perhaps I need to make a new feature request at the Github.

Sijun

2017년 8월 30일 수요일 오후 11시 13분 35초 UTC+9, mateusz.paprocki 님의 말:

Hi,

On Wed, Aug 30, 2017 at 9:53 AM, Sijun Cho [email protected] wrote:

I am trying to put up an extra y axis with log scale using the following code:

p = figure(plot_width=800, plot_height=400, toolbar_location=None)

p.line([1,2,3], [1,2,3])

p.y_range = Range1d(0,4)

p.circle([1,2,3],[10,100,1000], y_range_name=‘Y’)

p.extra_y_ranges[‘Y’]=Range1d(0,1000)

p.add_layout(LogAxis(y_range_name=‘Y’, axis_label=‘H’), ‘right’)

show§

When I use LinearAxis, the y axis is rendered well along with ticks and lables. However with LogAxis, only axis label is shown.

What goes wrong in the above code?

the problem is in lack of log scale on the y-axis for circle glyph. A scale is what’s actually responsible for mapping data to screen coordinates. It looks like currently it’s not possible to configure a custom scale along a custom range. I’m not sure right now if there is any inherent problem with such setup, or this is just an oversight.

Mateusz

You received this message because you are subscribed to the Google Groups “Bokeh Discussion - Public” group.

To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].

To post to this group, send email to [email protected].

To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/46adfb00-2a9f-41f5-a33f-cebb71afb974%40continuum.io.

For more options, visit https://groups.google.com/a/continuum.io/d/optout.

You received this message because you are subscribed to the Google Groups “Bokeh Discussion - Public” group.

To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].

To post to this group, send email to [email protected].

To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/7d60fab6-dd6b-4ac4-a125-245137a95850%40continuum.io.

For more options, visit https://groups.google.com/a/continuum.io/d/optout.

You received this message because you are subscribed to the Google Groups “Bokeh Discussion - Public” group.

To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].

To post to this group, send email to [email protected].

To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/644f2bbe-0da2-470e-870a-2bbf874403f8%40continuum.io.

For more options, visit https://groups.google.com/a/continuum.io/d/optout.

You received this message because you are subscribed to the Google Groups “Bokeh Discussion - Public” group.

To post to this group, send email to [email protected].

To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/0b076d0a-1eb8-47f7-b42c-6bd314cde509%40continuum.io.

For more options, visit https://groups.google.com/a/continuum.io/d/optout.

You received this message because you are subscribed to a topic in the Google Groups “Bokeh Discussion - Public” group.

To unsubscribe from this topic, visit https://groups.google.com/a/continuum.io/d/topic/bokeh/1tH2BXDxkdM/unsubscribe.

To unsubscribe from this group and all its topics, send an email to [email protected].

To post to this group, send email to [email protected].

For more options, visit https://groups.google.com/a/continuum.io/d/optout.

To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/bokeh/3624747A-6F30-4756-A2A2-C12B3EDED8D8%40anaconda.com.