Hide empty parts in a graph (dates)

Community Support
1

Hello, I have a hard time finding a solution to exclude some empty parts of my graph in bokeh 1.4.0.

First take a look at the graph:

image
image2197×734 48.4 KB

There is a large empty space between the two plots.

These are the interesting parts of my code:

source = ColumnDataSource(
        dict(x=x, y=y, label=benchmark, changeset=changeset,
             group_by=group_by, url_params=url_params))


p = figure(tools=TOOLS,
                x_axis_type='datetime',
                x_axis_label='Datetime',
                y_axis_type=y_axis_type,
                y_range=y_range,
                y_axis_label=y_axis_label,
                sizing_mode="stretch_both")

p.xaxis.formatter = DatetimeTickFormatter(
        milliseconds=['%Y-%m-%d %H:%M:%S.%3N'],
        seconds=["%Y-%m-%d %H:%M:%S"],
        minsec=["%Y-%m-%d %H:%M:%S"],
        minutes=["%Y-%m-%d %H:%M:%S"],
        hourmin=["%Y-%m-%d %H:%M:%S"],
        hours=["%Y-%m-%d %H:%M:%S"],
        days=["%Y-%m-%d %H:%M:%S"],
        months=["%Y-%m-%d %H:%M:%S"],
        years=["%Y-%m-%d %H:%M:%S"]
    )
p.xaxis.major_label_orientation = pi / 4

p.xaxis.major_label_overrides = {
            i: date.strftime('%Y-%m-%d %H:%M:%S') for i, date in enumerate(x)
        }

I have tried a similar approach to:
How do I make omit missing dates gaps in graph when using datetime as x-axis

but it doesn’t seem to work.

The data in p.xaxis.major_label_overrides are equal to:

{0: '2019-09-05 11:28:25', 1: '2019-09-05 11:28:25', 2: '2019-09-05 11:28:25', 3: '2019-09-05 11:28:26', 4: '2019-09-05 11:28:26', 5: '2019-09-05 11:28:26', 6: '2019-09-05 11:28:26', 7: '2019-09-05 11:28:26', 8: '2019-09-05 11:28:26', 9: '2019-09-05 11:28:26', 10: '2019-09-05 11:28:26', 11: '2019-09-05 11:28:26', 12: '2019-09-05 11:28:26', 13: '2019-09-05 11:28:26', 14: '2019-09-05 11:28:26', 15: '2019-09-05 11:28:26', 16: '2019-09-05 11:28:26', 17: '2019-09-05 11:28:26', 18: '2019-09-05 11:28:26', 19: '2019-09-05 11:28:26', 20: '2019-09-05 11:28:26', 21: '2019-09-05 11:28:26', 22: '2019-09-05 11:28:26', 23: '2019-09-05 11:28:26', 24: '2019-09-05 11:28:26', 25: '2019-09-05 11:28:26', 26: '2019-09-05 11:28:26', 27: '2019-09-05 11:28:26', 28: '2019-09-05 11:28:26', 29: '2019-09-05 11:28:26', 30: '2019-09-05 11:28:26', 31: '2019-09-05 11:28:27', 32: '2019-09-05 11:28:27', 33: '2019-09-05 11:28:27', 34: '2019-09-05 11:28:27', 35: '2019-09-05 11:28:27', 36: '2019-09-05 11:28:27', 37: '2019-09-05 11:28:27', 38: '2019-09-05 11:28:27', 39: '2019-09-05 11:28:27', 40: '2019-09-05 11:28:27', 41: '2019-09-05 11:28:27', 42: '2019-09-05 11:28:27', 43: '2019-09-05 11:28:27', 44: '2019-09-05 11:28:27', 45: '2019-09-05 11:28:27', 46: '2019-09-05 11:28:27', 47: '2019-09-05 11:28:27', 48: '2020-03-25 02:44:18', 49: '2020-03-26 02:44:40', 50: '2020-03-27 02:46:48', 51: '2020-03-28 02:45:58', 52: '2020-03-29 03:45:20', 53: '2020-03-30 02:47:09', 54: '2020-03-31 02:46:20', 55: '2020-04-01 02:48:40', 56: '2020-04-02 02:46:57', 57: '2020-04-03 02:45:12', 58: '2020-04-06 02:46:22', 59: '2020-04-07 02:45:21', 60: '2020-04-08 02:51:35', 61: '2020-04-09 02:45:40', 62: '2020-04-10 02:46:22', 63: '2020-04-13 02:45:26', 64: '2020-04-14 02:44:30', 65: '2020-04-15 02:45:55', 66: '2020-04-16 02:47:15', 67: '2020-04-17 02:46:08', 68: '2020-04-19 02:48:13', 69: '2020-04-20 02:46:27', 70: '2020-04-21 02:45:39', 71: '2020-04-22 02:47:06', 72: '2020-04-23 02:50:58', 73: '2020-04-24 02:47:09', 74: '2020-04-26 02:44:59', 75: '2020-04-27 02:50:34', 76: '2020-04-28 02:46:28', 77: '2020-04-29 02:46:34', 78: '2020-04-30 02:44:52'}
2

The linked solution relies on converting your x-coordinates to not be actual datetime values at all anymore. Instead they should be a simple integer row index (and there is no use for a DatetimeTickFormatter anymore, because the values are no longer datetime values).

1 Like
3

Thank you, I removed the DatetimeTickFormatter, but that doesn’t solve the problem. I think the format of the p.xaxis.major_label_overrides is correct, right?

Is there another approach to remove the empty space between the two parts of the plot?

4

@katamarandameko Did you convert the x-coordinates to be integer indices instead of datetimes? That is actually the crucial operation that has to be performed for the solution. Removing the ticker is not any part of the solution, I only mentioned it because the solution makes having that ticker be superfluous.