KPMG Visualisations (dash_utils)

• Version: 1.0.0
• Released: Juli 2019
• Web page: http://eskapade.kave.io
• Repository: https://github.com/kaveio/Eskapade-visualisations

This specific module is to create a data exploration and demo app using Dash (https://dash.plot.ly/). In addition, there is an interactive demo for the phi_k correlation analyzer (https://github.com/KaveIO/PhiK), which will calculate and display correlation matrices between two variables.

Check it out

To install check out the github repository:

    $ git clone https://github.com/KaveIO/Eskapade-visualisations
    $ pip install -e Eskapade-visualisations

Dash Template

Please see dash_tutorial.rst for more details on the tutorial and template.

Dash demo's

We created a couple of demos to showcase what you can do with dash and the dash_utils in the package.

df summary

Remix of the original df_summary link in Eskapade, now reworked to incorporate the pandas_profiling calculations as well. You can run the application by calling python [path/to/]df_summary.py which will then run on http://localhost:8050. You can select any variable and it will be plotted in either a barchart or a histogram, depending on the data type. In addition, the calculated values from the pandas_profiling will be shown in the table next to it.

TODO: Add a data uploader

template app

A template example using the functions from dash_utils, such as the figure_grid and the callback functions. It shows a basic example with 3 graphs and a table, that will respond to your selections. The app itself functions as an extension to the df_summary app. Where there you could only select a barchart, here you can also add a scatterplot and compare multiple variables.

TODO: Add a data uploader

The Phi_K demo

Demonstrates the actual capabilities of the phi_k correlation analyzer package. Consists of one file, phik_frontend.py which can be called from the heatmap folder using _python phik_frontend.py to start the flask server. NB: as it currently stands, if the app is called from outside the heatmap folder the .css files will not load correctly and the app will not be formatted correctly. As per default the app will server on http://localhost:8050. This version supports calculating correlation matrices for ordinal, categorical and continuous variables. It also allows for manual editing of the bin edges for continuous variables. The demo uses the diamond dataset as input.

Contact and support

Issues & Ideas: https://github.com/kaveio/Eskapade-visualisations/issues

Q&A Support: contact us at: kave [at] kpmg [dot] com

Please note that the KPMG Eskapade group provides support only on a best-effort basis.