mpl-markers

Interactive data markers for matplotlib.

Installation

pip install mpl-markers

Usage

import mpl_markers as mplm

Line Markers

Add a marker attached to matplotlib data lines:

import numpy as np
import matplotlib.pyplot as plt

fig, ax = plt.subplots(1,1)
x1 = np.linspace(-np.pi, np.pi, 1000)

ax.plot(x1, np.sin(x1)*np.cos(x1)**2)
# create line marker at x=0.
mplm.line_marker(x=0)

In interactive matplotlib backends (i.e. Qt5Agg), the marker can be dragged to any location along the data line, or moved incrementally with the left/right arrow keys.

The following keyboard shortcuts are supported in interactive mode, (note that the mouse must be inside the axes to register a keyboard event)

When moving markers interactively, the labels of the active marker may overlap with other markers on the axes. To stack overlapping markers, redraw the figure with F5.

Markers can be placed by x position, x/y position, or by the line data index. If index value(s) are given, they will override the x/y arguments if they are given.

# turn the label outline off, and use a rounded box style instead of the default square
mplm.set_style(
    ylabel=dict(fontfamily="monospace", bbox=dict(linewidth=0, facecolor="white", boxstyle="round4"))
)

fig, ax = plt.subplots(1,1)
x1 = np.linspace(-np.pi/2, np.pi/2, 1000)
ax.plot(x1, np.sin(x1) * np.cos(x1)**2)

# place markers every 100 points
mplm.line_marker(idx = np.arange(100, 1000, 100))

Axis Markers

Axis markers move freely on the canvas and are not attached to data lines. Axis markers can reference other markers to create a delta marker.

fig, ax = plt.subplots(1, 1)
y1 = np.random.normal(6, 3, size=10)

ax.bar(np.arange(10), y1)
ax.margins(x=0.2)

# create horizontal axis marker
m1 = mplm.axis_marker(y=np.min(y1), yformatter="{:.2f}%")

# create second marker that is referenced from the first marker m1
mplm.axis_marker(y=np.max(y1), ref_marker=m1, yformatter="{:.2f}%")

Meshgrid Markers

Data markers can also be added to pcolormesh plots. The marker label shows the value of the color-mapped z data.

xy = np.linspace(-1, 1, 100)
x, y = np.meshgrid(xy, xy)
z = np.sin(2*x)**2 + np.cos(3*y)**2

fig, ax = plt.subplots(1, 1)
m = ax.pcolormesh(x, y, z, vmin=0, vmax=2)
plt.colorbar(m)

# add a data marker at a single x/y point on the plot. x/y is in data coordinates.
mplm.mesh_marker(x=0.75, y=0)

Scatter Plot Marker

Markers attached to scatter plots can be added with scatter_marker.

data_x = np.random.normal(-1, 1, 100)
data_y = np.random.normal(-1, 1, 100)

fig, ax = plt.subplots(1, 1)
s = ax.scatter(data_x, data_y, color="b")

# place the marker on the point closest to x=0.5, y=0.5.
mplm.scatter_marker(
    x=0.5,
    y=0.5,
    yformatter=lambda x, y, pos: f"x={x:.2f}\ny={y:.2f}",
    scatterdot=dict(color="cyan", markeredgecolor="cyan"), # change the color of the active point
    anchor="upper left",
)

Markers in Jupyter Notebooks

Markers are non-interactive in juypter notebooks with the default matplotlib backend. To enable interactive markers, first install the ipympl backend,

pip install ipympl

Then enable the backend in jupyter with,

matplotlib.use("ipympl")

This backend is still experimental with mpl-markers, expect a few issues when using it. It does not support blitting (the image is entirely redrawn and exported as a png on each draw event), so there is some lag when dragging markers.

See this notebook for a full example.

Styling

The marker style is controlled by the mpl_markers/style/default.json file:

{
    "xline": {
        "linewidth": 0.6,
        "color": "k",
        "linestyle": "dashed"
    },
    "yline": {
        "linewidth": 0.6,
        "color": "k",
        "linestyle": "dashed"
    },
    "xlabel": {
        "fontsize": 8,
        "color": "black",
        "padding": 5,
        "bbox": {
            "boxstyle": "square",
            "facecolor": "white",
            "edgecolor": "black",
            "alpha": 1,
            "linewidth": 1.5
        }
    },
    "ylabel": {
        "fontsize": 8,
        "padding": 5,
        "offset": 10,
        "bbox": {
            "boxstyle": "square",
            "facecolor": "white",
            "edgecolor": "black",
            "alpha": 1,
            "linewidth": 1.5
        }
    },
    "zlabel": {
        "fontsize": 8,
        "padding": 10,
        "bbox": {
            "boxstyle": "square",
            "facecolor": "white",
            "edgecolor": "black",
            "alpha": 1,
            "linewidth": 1.5
        }
    },
    "datadot": {
        "markersize": 10,
        "marker": "."
    },
    "axisdot": {
        "markersize": 10,
        "marker": ".",
        "markerfacecolor":"white", 
        "markeredgecolor":"k"
    },
    "scatterdot": {
        "markersize": 10,
        "marker": ".",
        "markeredgewidth": 1,
        "markeredgecolor": "white"
    }
}

To use custom styles, pass in a dictionary of artist settings to the set_style method. Keys that do not match those found in default.json are ignored. Settings are applied globally to all future markers.

mplm.set_style(
    ylabel=dict(fontfamily="monospace", bbox=dict(linewidth=0, facecolor="none"))
)

To limit the style scope to a specific axes, artist settings can also be passed into init_axes as kwargs. Settings can also be passed to individual markers when they are created, and allows for mixed styles on the same axes.

# Inherits global settings from the last set_style call.
mplm.line_marker(
    x=np.pi/4,
    ylabel=dict(fontsize=11),             
    xline=False # turn off the xline artist
)

Custom .json files are supported with the mplm.set_style_json method. This will set the style on all future markers and must have the same keys as the default.json file:

mplm.set_style_json("user_style.json")

License

mpl-markers is licensed under the MIT License.