imagesc

• imagesc is an Python package to create heatmaps. Various methods to create a heatmap are implemented, each with specific properties that can help to easily create your heatmap. The fast and clean method is optimized for speed, the cluster method provides clustering, the seaborn method contains many configuration settings, and finally, the plot as good as possible the imagesc from matlab.

Functions in imagesc

# X is your numpy array
fig = imagesc.seaborn(X)
fig = imagesc.cluster(X)
fig = imagesc.fast(X)
fig = imagesc.clean(X)
fig = imagesc.plot(X)
status = imagesc.savefig(fig)
out = imagesc.d3(X)

Contents

• Installation
• Requirements
• Quick Start
• Contribute
• Citation
• Maintainers
• License

Installation

• Install imagesc from PyPI (recommended). imagesc is compatible with Python 3.6+ and runs on Linux, MacOS X and Windows.
• It is distributed under the MIT license.

Requirements

# Note that: seaborn is only required when using **seaborn** or **cluster** functions.
pip install -r requirements.txt

Installation from Pypi

pip install imagesc

Import imagesc package

import imagesc as imagesc

d3 example

• Implemention is based on d3
• Interactive
• https://d3-graph-gallery.com

import imagesc
df = pd.DataFrame(np.random.randint(0, 10, size=(52, 7)))
paths = imagesc.d3(df, fontsize=10, title='Hooray!', description='python to d3 conversion using imagesc!', path='d3heatmap.html', width=750, height=200, cmap='interpolateGreens')

seaborn

• Underlying implemented is based on seaborn
• Large number of configurations
• Slow when using large datasets
• Grid is aligned to the cells
• See here for all parameters: https://seaborn.pydata.org/generated/seaborn.heatmap.html

df = pd.DataFrame(np.random.randint(0,100,size=(10,20)))
A = imagesc.seaborn(df.values, df.index.values, df.columns.values)
B = imagesc.seaborn(df.values, df.index.values, df.columns.values, annot=True, annot_kws={"size": 12})
C = imagesc.seaborn(df.values, df.index.values, df.columns.values, annot=True, annot_kws={"size": 12}, cmap='rainbow')
D = imagesc.seaborn(df.values, df.index.values, df.columns.values, annot=True, annot_kws={"size": 12}, cmap='rainbow', linecolor='#ffffff')

A B C D

cluster

• Underlying implemented is based on clustermap
• When you desire to cluster your heatmap
• Default distance setting: metric="euclidean", linkage="ward" (can be changed)
• Slow for large data sets
• Grid is aligned to the cells
• Possibilities to tweak
• Possible arguments: https://seaborn.pydata.org/generated/seaborn.clustermap.html

df = pd.DataFrame(np.random.randint(0,100,size=(10,20)))
fig_C1 = imagesc.cluster(df.values, df.index.values, df.columns.values)
fig_C2 = imagesc.cluster(df.values, df.index.values, df.columns.values, cmap='rainbow')
fig_C3 = imagesc.cluster(df.values, df.index.values, df.columns.values, cmap='rainbow', linecolor='#ffffff')
fig_C4 = imagesc.cluster(df.values, df.index.values, df.columns.values, cmap='rainbow', linecolor='#ffffff', linewidth=0)
imagesc.savefig(fig_C1, './docs/figs/cluster4.png')

C1 C2 C3 C4

fast

• Underlying implemented is based on pcolorfast
• Fast
• Not so much tweakable
• Grid is not aligned to the cells
• Possible arguments: https://matplotlib.org/3.1.1/api/_as_gen/matplotlib.axes.Axes.pcolorfast.html

df = pd.DataFrame(np.random.randint(0,100,size=(10,20)))
fig_F1 = imagesc.fast(df.values, df.index.values, df.columns.values)
fig_F2 = imagesc.fast(df.values, df.index.values, df.columns.values, grid=False)
fig_F3 = imagesc.fast(df.values, df.index.values, df.columns.values, grid=False, cbar=False)
fig_F4 = imagesc.fast(df.values, df.index.values, df.columns.values, grid=True, cbar=False)
fig_F5 = imagesc.fast(df.values, df.index.values, df.columns.values, cmap='rainbow')
fig_F6 = imagesc.fast(df.values, df.index.values, df.columns.values, cmap='rainbow', linewidth=0.5, grid=True)
imagesc.savefig(fig_C1, './docs/figs/fast1.png')

F1 F2 F3 F4 F5 F6

clean

• Underlying implemented is based on pcolorfast
• Fast
• No Grid
• Limited configurations
• Ideal for photos
• Possible arguments: https://matplotlib.org/3.1.1/api/_as_gen/matplotlib.axes.Axes.pcolorfast.html

df = pd.DataFrame(np.random.randint(0,100,size=(10,20)))
fig_FC1 = imagesc.clean(df.values)
fig_FC2 = imagesc.clean(df.values, cmap='rainbow')
imagesc.savefig(fig_C1, './docs/figs/clean1.png')

F1 F2

plot

• Underlying implemented is based on imshow
• implementation will behave more-or-less as the one of matlab
• Medium speed
• Various configurations are possible but less then seaborn
• Grid is aligned to the cells
• Possible arguments: https://matplotlib.org/3.1.1/api/_as_gen/matplotlib.pyplot.imshow.html

df = pd.DataFrame(np.random.randint(0,100,size=(10,20)))
fig_M1 = imagesc.plot(df.values)
fig_M2 = imagesc.plot(df.values, cbar=False)
fig_M3 = imagesc.plot(df.values, cbar=False, axis=False)
fig_M4 = imagesc.plot(df.values, cbar=False, axis=True, linewidth=0.2)
fig_M5 = imagesc.plot(df.values, df.index.values, df.columns.values)
fig_M6 = imagesc.plot(df.values, df.index.values, df.columns.values, cbar=False, linewidth=0.2)
fig_M7 = imagesc.plot(df.values, df.index.values, df.columns.values, grid=True, cbar=False, linewidth=0.2)
fig_M8 = imagesc.plot(df.values, df.index.values, df.columns.values, grid=False, cbar=False, linewidth=0.2)
fig_M9 = imagesc.plot(df.values, df.index.values, df.columns.values, grid=True, cbar=False, linewidth=0.8, linecolor='#ffffff')
fig_M10 = imagesc.plot(df.values, df.index.values, df.columns.values, grid=True, cbar=False, linewidth=0.8, linecolor='#ffffff', cmap='rainbow')
imagesc.savefig(fig, './docs/figs/plot10.png')imagesc.savefig(fig_C1, './docs/figs/fast1.png')

M1 M2 M3 M4 M5 M6 M7 M8 M9 M10

Speed:

import matplotlib.image as mpimg
img=mpimg.imread('./docs/figs/lenna.png')

fig = imagesc.clean(img)
# runtime 1.49

fig = imagesc.fast(img, cbar=False, axis=False)
# runtime: 2.931 seconds

fig = imagesc.plot(img, linewidth=0, cbar=False)
# runtime: 11.042

**fast** **clean** **plot**

Citation

Please cite imagesc in your publications if this is useful for your research. Here is an example BibTeX entry:

@misc{erdogant2019imagesc,
  title={imagesc},
  author={Erdogan Taskesen},
  year={2019},
  howpublished={\url{https://github.com/erdogant/imagesc}},
}

References

• seaborn https://seaborn.pydata.org/generated/seaborn.heatmap.html
• clustermap https://seaborn.pydata.org/generated/seaborn.clustermap.html
• fast and clean https://matplotlib.org/3.1.1/api/_as_gen/matplotlib.axes.Axes.pcolor.html
• plot https://matplotlib.org/3.1.1/api/_as_gen/matplotlib.pyplot.imshow.html
• Other https://matplotlib.org/3.1.1/gallery/images_contours_and_fields/image_annotated_heatmap.html
• Colormap https://matplotlib.org/examples/color/colormaps_reference.html

Maintainer

• Erdogan Taskesen, github: erdogant
• Contributions are welcome.
• If you wish to buy me a Coffee for this work, it is very appreciated :)