Evgeniy Potapov · Jun 3 3m read

Water Conditions in Europe app with Python and Dashboards

It is very interesting to compare different BI technologies. It is curious to me what the differences are in functionality, development tools, speed and usability.

For this application, I chose a dataset with water conditions in various European countries. This is an open source dataset containing observational data from 1991 to 2017.

The team and I decided to make a model based on this BI dataset using IRIS BI, Tableau, PowerBI and InterSystems Reports (powered by Logi Reports).

For the frontend, we made a web interface in PythonFlask via Embedded Python.

The result, by the way, can be seen on the page
You can look at the demo stand, because deploying a container from the repository can take up to 20 minutes. Lots of python packages, more on why later.

Main page


Actually the data seemed small, the period is only 17 years :) . 

Therefore, in addition to the existing ones, I wanted to continue the data set and used a neural network for this. Using the same Embedded Python, used Tensorflow, this package occupies 511MB when downloaded, don't be surprised.

Actually, this is the reason for the long deployment of containers - a lot of packages are downloaded for the neural network, quite a lot of related packages and a long installation. But there will be a separate article about neural networks and Integrated ML, I will publish it soon.

I will also say that the result of the prediction is entered into the same database, so you can see the dataset through BI tools. But predict was made only for one river in France, take a closer look. Because there was enough power to calculate only one thing. The full prediction took a very long time.


Data cubes are made in IRIS and also cubes are made in Adaptive Analytics (powered by AtScale). Thus, IRIS BI dashboards were built on IRIS, and the rest of the tools (Logi, PowerBI, Tableau) took data from AtScale. 


Actually BI systems are presented in a variety.

This is Dash in python.

Our Favorite IRIS BI

InterSystems Reports (powered by Logi Report)


Tableau Example


For all the BI systems presented, source files are available in the repository. You can see how the reports and dashboards work and use them for your future projects.

On them you can visually see how a particular system works, as well as understand what development tools are available in a particular system.

There was a little lack of time to make a page where you could enter various parameters and conditions in order to get a calculation of the likely value using a neural network. There are all the prerequisites for this, an example of the implementation of a neural network and predictions with its help, which is implemented in Embeded Python. The neural network is trained, startup scripts are available, located in the folder

Thank you for reading the article, I'm waiting comments and feedback. Soon there will be an article about comparing a neural network in Embedded Python and IntegratedML, I will attach it to this article.

0 109
Discussion (0)1
Log in or sign up to continue