Polirural Lab is a platform to facilitate experimentation and data science oriented tasks as well as publishing prototype applications has been developed by BOSC. We call it DIH Lab in this document. It showcases system dynamics models and in future will also provide running it with different parameters.

The main building blocks of the DIH are:

• Dashboard
• External apps (mainly maps)
• List of datasets with descriptions
• Workspaces
• Models (System dynamics)
• Maps for calculating and displaying of rural attractiveness and clusterization of regions

The DIH Lab is a platform where users can have their own isolated docker based  environment to perform data science related tasks involving data mining, processing and visualization.

 

To provide this functionality JupyterHub platform has been tightly integrated into the DIH Lab by developing user account synchronization, Jupyter notebook transfering, execution using Papermill and chaining of outputs and visualizing them using one or a combination of methods:

• Maps based on Hslayers-ng mapping library
• Data tables 
• Images
• Plain text output

A visual modelling tool has been developed to make the process of chaining the building blocks of data-processing-pipeline easier. Linking datasets to processing nodes, generates code in the Jupyter notebook which is behind each processing node. (Figure 7)

Currently it is possible to run notebooks which import the most common data science related libraries, such as geojson, datapackage, pandas, plotly, geopandas, chart-studio, matplotlib, descartes, scikit-learn, but further extension of this list will be done after initial feedback from the users. Also more work will be needed to support the various output types it is possible to generate using jupyter notebooks and mapping output to visualization components.

Applications can be also developed outside this platform and deployed later by providing a git repository URL. The applications run inside an isolated docker container behind a reverse proxy provided by the Hub infrastructure. This way it will be possible to use the vast amount of data on the server without needing to download it on developers machines.

 

Explore the Polirural Lab at  https://dih.bosc.lv/