Articles
Cosmo Tech provides a 360° Simulation Digital Twin platform that allows users to build and deploy digital models of their systems and explore multiple scenarios to solve the most complex industrial problems and lead enterprise decision making. Companies from a wide range of industries rely on Cosmo Tech to predict the evolution of their organization to better understand the impact of their decisions and optimize all levels of enterprise planning, ensuring a future that is robust, resilient and sustainable.
Our platform is powered by Azure which makes it easy to integrate and benefit from services in the Microsoft Azure ecosystem, including our topic for today: the Azure Machine Learning studio. The Azure Machine Learning studio is the web portal for data scientist developers in Azure Machine Learning. The studio combines no-code and code-first experiences for an inclusive data science platform. You can write and run your own code in managed Jupyter Notebook servers that are directly integrated in the studio, train and deploy machine learning models with Azure Machine Learning designer, create automated ML experiments with an easy-to-use interface, and use Azure Machine Learning data labeling to efficiently coordinate image labeling or text labeling projects (see What is Azure Machine Learning studio?)
In this article, we will discuss how to combine these two powerful solutions together; three use cases will be presented:
Without further ado, let’s start!
For a typical Cosmo Tech solution, we offer a web application to manage digital twin simulations. You can create a scenario, upload your dataset, configure and launch a simulation – all with an intuitive graphical interface designed for business users.
However, for an advanced user, this graphical interface may be unnecessary – perhaps you want to do all the operations above with the command line so that you can do it automatically and regularly. For example, you have daily updated data and you want to test it every day.
We provide an open source Python client for the Cosmo Tech Platform API (you can find it on Github here: https://github.com/Cosmo-Tech/cosmotech-api-python-client). In the AzureML studio you have access to the compute instance terminal. You can easily install the Python client with PIP with following command in the terminal:
pip install git+https://github.com/Cosmo-Tech/cosmotech-api-python-client.git
Once it’s installed, you can launch all kinds of services provided by the Cosmo Tech Platform API. AzureML studio allows users to work with files, folders and Jupyter Notebooks directly in the workspace. For example, you can run a simulation with following Python code in a Jupyter Notebook:
This client works in any environment with Python installed. The advantages of using AzureML studio are:
Data visualization is essential for data science projects, it helps us understand data better. We embed Power BI reports in our web application for data visualization. However, for advanced users, if you want to create your own customized data visualizations, it’s also possible to visualize Cosmo Tech digital twin simulation results in Jupyter Notebooks in the AzureML studio.
First, you need to connect to the database. Cosmo Tech uses ADX (Azure Data Explorer) for our simulation data storage. It is a fully-managed big data analytics cloud platform and data-exploration service. In order to explore data from ADX, you can install KustoClient for Python:
pip install azure-kusto-data
Then use the following code to connect to an ADX database:
Once you are connected to the database, you can run Kusto queries to process data and return results as a Pandas dataframe:
Then you can use any preferred Python visualization library to create charts.
Data is the most important and fundamental element for machine learning. The quality and quantity of data can largely decide the performance of your machine learning model. However in many domains it’s both difficult and expensive to collect data.
Cosmo Tech digital twin platform offers you the tools to build and deploy simulable digital twins of your system, which are accurate digital representations of the structure and processes of your system. The platform allows you to run simulations on these models, virtual experimentations that cannot be done with the real system. By running simulations with the digital twins, we can generate data efficiently. In this way, the Cosmo Tech digital twin platform can serve as a data source for machine learning.
There’s no doubt that it’s challenging to combine these two cutting edge technologies. If you are interested in the concept of integrating digital twins with machine learning, don’t hesitate to contact us for further discussion.
In this article, we present three use cases to combine the Cosmo Tech digital twin platform with Azure machine learning studio: Cosmo Tech digital twin simulations in AzureML studio, Customized data visualization of Cosmo Tech digital twin simulations in AzureML studio, and Machine learning based on Cosmo Tech digital twin simulations in AzureML studio. You can use AzureML studio to have more flexibility and customization for the Cosmo Tech digital twin platform and use Cosmo Tech digital twin platform to generate data for AzureML studio. Thanks for reading!