Negative scatter selection

I am getting peculiar behavior when trying to use BoxSelectTool. When I go to select points, it fails to capture the negative points. (1st and 2nd plot)

I determined that this is not specific to the separate renders. If I shift all of the data up, the positive points in both renders are able to be selected. (3rd and 4th plots)

If I need to, I can workaround this problem by adding a large offset and then relabeling the y-tick labels, but I’d rather fix this the proper way if there is some setting that I’m missing.

I’m not sure if this is related, but I also noticed that if I don’t specify the y_range, it will only show positive values. It will scale the y-axis to accommodate whatever positive values I give it, but it does not adjust to contain any negative values. I was able to to view the negative values only after specifying y_range = [-6,6]

@alustig3 it is impossible to speculate without a complete
Minimal Reproducible Example and version information (which should always be specified).

Sorry for lack of information. In the process of reducing my code to a minimal reproducible example…I discovered the problem. It’s almost as if creating a minimal reproducible example not only benefits those trying to help, but is also is a good way to debug things in the first place and may prevent the forum from getting spammed :grin:

In case future users stumble upon this post with similar difficulties, the problem was that my x-values were a list of numeric strings instead of a list of actual numbers.

Here is a .gif showing the resulting problem of negative numbers not being selected in the first plot. The second plot shows the expected behavior that happens when properly using numbers for the x-values.
CleanShot 2022-01-08 at 03.24.48
Below is the code that demonstrates the problem and the solution (shown in the video). I am using Bokeh 2.4.2 with Python 3.10.0

import bokeh.plotting as bk
from bokeh.models import BoxSelectTool,ColumnDataSource
from bokeh.layouts import column

trial_strings = ['1','2','3','4','5','6','7','8','9']
trial_nums = [1,2,3,4,5,6,7,8,9]
vals =   [1,2,1,1,-1,-2,-1,1,1]

bad_source  = ColumnDataSource(
          'vals': vals }
good_source  = ColumnDataSource(
          'vals': vals }

bad_plot = bk.figure(
    width = 800,
    height = 250,
    tools = "reset",
    title = 'Bad plot. If using strings for xvals, you cannot select negative yvals',
    y_range = [-6,6],
good_plot = bk.figure(
    width = 800,
    height = 250,
    tools = "reset",
    title = 'Working plot',
    y_range = [-6,6],

bad_render = bad_plot.square('trials','vals',source=bad_source, size=12)
good_render = good_plot.square('trials','vals',source=good_source, size=12)


@Bryan thank you for your quick response and for creating such a wonderful tool!

1 Like

@alustig3 Glad you found a solution!