Inspector plot#

The inspector plot takes in a three-dimensional input and applies a reduction operation (sum by default) along one of the dimensions specified by dim. It displays the result as a two-dimensional image.

In addition, the inspector tool 737f65ef2537478e85429e81db24abf3 in the toolbar allows to add markers on the image, which generate one-dimensional slices (retaining only dim) that are displayed in a second figure.

The points can be dragged/moved, as well as deleted (via a middle click). Below is an example on how to create the inspector plot.

[1]:

%matplotlib widget
import plopp as pp
from plopp.data.examples import clusters3d

[2]:

da = clusters3d(nclusters=50, seed=12).hist(z=100, y=100, x=100)

p = pp.inspector(da, dim='z', orientation='vertical', logc=True)

[4]:

[4]:

Controls

Left-click to make new point
Left-click and drag existing point to move it
Middle-click to delete point

Rectangle mode#

Instead of using dots to select a single pixel in the 2D image, we can enable 'rectangle' mode to draw rectangles that select everything contained by them, using the 55b85ac89da24485a1cfeea21a63bdfe tool.

[5]:

p = pp.inspector(da, dim='z', orientation='vertical', logc=True, mode='rectangle')

[7]:

[7]:

Controls

Left-click to start drawing a new rectangle
Left-click again to persist the rectangle
Left-click and drag a vertex to resize the rectangle
Right-click and drag to move the entire rectangle
Middle-click to delete the rectangle

Polygon mode#

Finally, there is also a 'polygon' d9a663ea10ed41c298c803643fccfef2 mode:

[8]:

p = pp.inspector(da, dim='z', orientation='vertical', logc=True, mode='polygon')

[10]:

[10]:

Controls

Left-click to start drawing a new polygon
Left-click again to add more points
Left-click and drag a vertex to resize the polygon
Right-click and drag to move the entire polygon
Middle-click to delete the polygon

Disabling continuous updates#

For performance reasons, it is sometimes preferable to only update the target figure when we have stopped dragging or resizing one of the rectangles/polygons, instead of every step of the way.

This can be done using the continuous_update argument:

[11]:

p = pp.inspector(
    da,
    dim='z',
    orientation='vertical',
    logc=True,
    mode='rectangle',
    continuous_update=False,
)

[13]:

[13]:

Changing the reduction operation#

As mentioned above, the default operation applied along the third dimension is a sum. This can be changed via the operation argument (possible other choices are mean, min, and max, as well as their nan-prefixed counterparts).

They can be used when summing the data along the third dimension does not really make physical sense. For instance, below we load a dataset containing the air-temperature recorded above North-America for 2 years. Summing temperatures along the time dimension to create a map would be illogical, and a much better representation of the data would be to plot the average temperature at a given location, over the 2 years measured.

[14]:

from plopp.data.examples import air_temperature

air = air_temperature()
air

Downloading file 'air_temperature.h5' from 'https://public.esss.dk/groups/scipp/plopp/1/air_temperature.h5' to '/home/runner/.cache/plopp/1'.

[15]:

inspect_air = pp.inspector(air, dim='time', orientation='vertical', operation='mean')

[17]:

inspect_air

[17]:

This Page

Inspector plot#

Rectangle mode#

Polygon mode#

Disabling continuous updates#

Changing the reduction operation#