hishiryo

Consciousness beyond thought ^_^

Hishiryo is a tool to generate a circular dataviz of any csv dataset.

This little experiment aims at trying to quickly represent the content of a dataset and make it funny to discover it's patterns!

Each datapoint (like a cell in an excel sheet) is converted into a pixel , and this pixel is diplayed on a circular graph.

This version supports the following column data formats : float, integers and text as nominal categories.

How to install

With PyPI (Make sure your pip is up to date as it can interfere with CV2 package install)

pip install hishiryo

Dependencies

• Pandas
• CV2 (OpenCV)
• Pillow
• svgwrite

How to use radial rendering

1 - find an appropriate csv. (more than 100 rows will start to overload the result a little)

2 - From your code or from a Jupyter notebook :

from hishiryo import Hishiryo

HishiryoConverter = Hishiryo.Hishiryo()

input_path = "/home/user/iris.csv"
separator = ','
output_path = "/home/user/iris.png"
radius = 500

HishiryoConverter.convertCSVToRadialBitmap(input_path,separator,output_path,radius,None,"Dot")

Arguments :

• input path is the path to your csv file (e.g. /home/user/iris.csv)

• output path is the path to your target image file (e.g. /home/user/iris.png) The fileformat you want is autodetected thanks to CV2 functionalities.

• separator is the character separator in your csv (e.g. ",")

• radius (in pixel) is the size of the radius of the disk where the pixels will be drawn. The higher it is the bigger and sharper your output image will be. (e.g: 1500)'

• sort_by is the name of the column or the list of column you want to sort you data. (e.g. "Sepal.Length", or ["Sepal.Length","Sepal.Width"])

• glyph_type is the type of representation you want for the pixels. it can be one among the following : "Dot","Square" or "Polygon"

    input_path = "path/to/your/csv/file.csv"
    output_path = "path/to/your/rendered/image.png"
    separator = ","
    radius = 3000
    sort_by = None
    glyph_type = "Polygon"
  

Colors are assigned based on the variable type.

• Blue circles represent integer values
• Red circles, float values
• Random colors are assigned to categorical variables.
• Black is the default value for value 0 or when there is no data

Radial bitmap rendering examples

See below an example of a visualisation generated from the train titanic dataset

The 12 columns of the dataset are represented as 12 circular rows. The first column is rendered as the inner circle, and the last as the outer one.

Here is an overview of the MNIST train dataset, sort by label

How to use simple bitmap rendering

Another way is to render your csv as a simple bitmap, without radial representation.

from hishiryo import Hishiryo

HishiryoConverter = Hishiryo.Hishiryo()

input_path = "/home/user/iris.csv"
separator = ','
output_path = "/home/user/iris.png"

HishiryoConverter.convertCSVToBitmap(input_path,
                                     output_path,
                                     separator,
                                     )

simple bitmap rendering example

Here is an extract of the MNIST test dataset ()

Licence

GNU General Public License v3.0