freeplot

FreePlot is a Python data visualization library based on Matplotlib. It provides some simple implementations according to my preference. Matplotlib is powerful yet not easy to draw what you want due to its complicated arguments. I feel FreePlot is more friendly, especially for papers.

Installation

pip install freeplot

Quick Recipe

• import

from freeplot.base import FreePlot

• create a new container

fp = FreePlot(shape=(2, 2), titles=('a', 'b', 'c', 'd'), sharey=True)

shape: the arrangement of axes, 2 x 2, a total 4 axes in this case;

figsize: (height, width), the real figure size here is 400 x 440;

titles: the title for each ax;

sharey: axes will share the y axis if true.

the container can be used in a ndarray-style, e.g.:

# fp[0, 0], fp[0, 1]

You can also use title:

# fp['a']

But the slice operation is not supported:

# fp[:, 0]

• plotting

I implement some methods for plotting such as lineplot, scatterplot ...

fp.lineplot(x=[1, 2, 3], y=[4, 5, 6], index=(0, 0), label='test')

• set xlabel, ylabel

use

fp.set_label('X', axis='x', index=(0, 0))
fp.set_label('Y', axis='y', index=(0, 0))

or

fp.set(xlabel='X', ylabel='Y', index=(0, 0))

• set title

fp.set_title(y=0.98) # for all axes

• save your fig

fp.savefig('test.pdf')

Example

Line, Scatter, Bar, Heatmap

Let's do a little complicate plotting.

import numpy as np
import pandas as pd
from freeplot.base import FreePlot



titles = ('Line', 'Scatter', 'Bar', 'Heatmap')
fp = FreePlot(shape=(2, 2), titles=titles, sharey=False)

# Line
x = np.linspace(0, 2, 10)
y1 = x ** 0.5
y2 = x ** 2
fp.lineplot(x, y1, index=(0, 0), style='line', label='sqrt(2)')
fp.lineplot(x, y2, index=(0, 0), style='line', label='pow(2)')
fp[0, 0].legend()

# scatter
x = np.random.randn(100)
y = np.random.randn(100)
fp.scatterplot(x, y, index='Scatter', style='scatter')
fp.set_label('X', index=(0, 1), axis='x')
fp.set(ylabel='Y', index=(0, 1))

# bar
A = [1., 2., 3.]
B = [2., 3., 4.]
T = ['One', 'Two', 'Three'] * 2
Hue = ['A'] * len(A) + ['B'] * len(B)

data = pd.DataFrame(
    {
        "T": T,
        "val": A + B,
        "category": Hue
    }
)
fp.barplot(x='T', y='val', hue='category', data=data, index=(1, 0), auto_fmt=True)

# Heatmap
row_labels = ('c', 'u', 't', 'e')
col_labels = ('l', 'r', 'i', 'g')
data = np.random.rand(4, 4)
df = pd.DataFrame(data, index=col_labels, columns=row_labels)
fp.heatmap(df, index='Heatmap', annot=True, fmt=".4f", cbar=False, linewidth=0.5)

# set titles
fp.set_title(y=0.98)

# savefig
fp.savefig('demo.png')
# fp.show()

Stack

import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
from freeplot.base import FreePlot

x = np.arange(0, 10, 2)
ay = [1, 1.25, 2, 2.75, 3]
by = [1, 1, 1, 1, 1]
cy = [2, 1, 2, 1, 2]
y = np.vstack([ay, by, cy])
fp = FreePlot()
fp.stackplot(x, y, index=(0, 0), style='stack')
fp.show()

Radar

import numpy as np
from freeplot.base import FreePlot
from freeplot.zoo import pre_radar, pos_radar



labels = (
    "brightness", "fog", "gaussian_blur", "glass_blur", "jpeg_compression",
    "motion_blur", "saturate, snow", "speckle_noise", "contrast", "elastic_transform", "frost",
    "gaussian_noise", "impulse_noise", "pixelate", "shot_noise", "spatter", "zoom_blur", "transform", "flowSong"
)

theta = pre_radar(len(labels), frame="polygon")

# shape: 1, 1; figsize: 4, 4;
fp = FreePlot((1, 1), (4, 4), dpi=100, titles=["RADAR"], projection="radar")
fp.set_style('no-latex')

data = {
    "A": np.random.rand(len(labels)),
    'B': np.random.rand(len(labels)),
    'C': np.random.rand(len(labels))
}

pos_radar(data, labels, fp, index=(0, 0), style='bright')

fp[0, 0].legend()

fp.savefig("radar.png", tight_layout=True)

Violin

import numpy as np
import matplotlib.pyplot as plt
from freeplot.base import FreePlot


fp = FreePlot((1, 1), (5, 5))
# note that each element is a group of data ...
all_data = [np.random.normal(0, std, 100) for std in range(5, 10)]
fp.violinplot(x=None, y=all_data, index=(0, 0))

fp.savefig('violin.png')

Inset_axes

from freeplot.base import FreePlot


fp = FreePlot((1, 1), (4, 5))

fp.lineplot([1, 2, 3], [4, 5, 6], label='a')
fp.lineplot([1, 2, 3], [3, 5, 7], label='b')
axins, patch, lines = fp.inset_axes(
    xlims=(1.9, 2.1),
    ylims=(4.9, 5.1),
    bounds=(0.1, 0.7, 0.2, 0.2),
    index=(0, 0),
    style='line' # !!!
)
fp.lineplot([1, 2, 3], [4, 5, 6], index=axins)
fp.lineplot([1, 2, 3], [3, 5, 7], index=axins)
fp.savefig('inset.png')

3D Surface

import numpy as np
import matplotlib.pyplot as plt
from freeplot.base import FreePlot


X = np.arange(-5, 5, 0.25)
Y = np.arange(-5, 5, 0.25)
X, Y = np.meshgrid(X, Y)
R = np.sqrt(X**2 + Y**2)
Z = np.sin(R)

fp = FreePlot(projection='3d', dpi=300) # projection: '3d' !!!
fp.surfaceplot(X, Y, Z, cmap=plt.cm.coolwarm, antialiased=False, linewidth=0)
fp.set_label(r"$x$", axis='x')
fp.set_label(r"$y$", axis='y')
fp.set_label(r"$z$", axis='z')
fp.show()

Latex

We turn off the 'latex' option by default to speed up. You can activate it by (make sure that Latex is accurately installed on your computer)

fp = FreePlot(latex=True)

You shall use the following to close it halfway,

fp.set_style('no-latex')

or

plt.style.use('no-latex')

Tips

• For lineplot, barplot ..., you can directly use matplotlib.axes._axes.Axes as index, e.g.:

  fp.lineplot(x, y, index=fp[0, 0])
  

• You may find some interesting implementations in freeplot.zoo, such as tsne, roc_curve ...

• You can use a list of styles:

  fp.set_style(['bright', 'high-vis', {"axes.facecolor":".9"}])
  

• It would be better to use

  fp.legend
  

  to set legend instance instead of

  fp[0, 0].legend
  

• You can find all available styles by

  fp.styles
  

• You can scale x|y-axis by

  fp.set_scale(value='symlog', index=(0, 0), axis='x')
  fp.set_scale(value='symlog', index=(0, 0), axis='y')
  

• You can add text on Axes by

  fp.set_text(x, y, s, fontsize=10)
  

• You can create Patches by FreePatches and add them by

  fp.add_patch(patch)