While the world is enjoying summer vacations, Datrics breaks limits to what’s possible with no-code analytics and data science platform. Let’s dive into custom code updates and new features in charts.
Datrics continues working on the custom code features in the platform. We aim to provide the right tooling for data scientist to use their expertise and approach complex data analytics tasks and machine learning without being restricted by the platform. New low code functionality provides all the benefits of no-code analytics platform, and keeps the freedom to experiment.
With the latest update Datrics removes the limits of the built-in Python libraries, giving the opportunity to load additional resources from pypi.
Find out more on how to add custom code bricks in Datrics.
There’s a new tab “Libraries” in the Custom code brick editor. To use the additional resource in your custom code brick you have to install the library from pypi repository. Pypi has a huge repository of the packages to use for machine learning, artificial intelligence or ETL data. Define name and version of the library and press Add. In case version is not provided, the latest will be installed.
Before adding the library to the list, we are validating the installation. There’s the result in the logs and the notification. Library may be successfully installed with warning or errors, on case there are no critical compatibility issues. One does not have to install the libraries that come out-of-the-box with Datrics. Check out the built-in Python libraries.
Once custom code updates are made, resources added, code written, we are saving the brick and finally install all the libraries to insure there are no problems for further pipeline runs. Depending on the amount of the libraries and their size, it may take some time. Be patient, it’s going to be worth it!
Custom brick may be used in any Datrics pipeline within the project it belongs to. There’s no limitation on compatibility of the packages among custom code bricks in the same pipeline. Libraries are installed for the particular custom brick, which lets Datrics run the pipeline step by step avoiding the conflicts that would occur otherwise. For example, you may use in the same pipeline 2 custom code bricks with the different versions of the same library, which otherwise would be not possible.
Now the custom code brick can also be copied as a separate asset or together with the pipeline.To copy brick as the separate asset, find it in the list in the left panel, in the context menu select “Copy to”. Select current or different project and set the new name for the brick.
The name of the custom code brick in the project should be unique. This will definitely help to find it easier afterwards. We are installing the libraries for the custom code brick, while it’s being copied, so that it’s ready to be used in the pipeline.
User may also copy the entire pipeline with custom bricks. When you are copying your pipeline to the different project, all the bricks will be copied as well. To run the pipeline you need to install the libraries manually for the newly created bricks.
For more details and instructions go to our wiki.
Visualizations is one of the key elements to data analysis. Therefore we continue improving the chart brick in our no-code analytics and data science tool.
Some advancements added:
To optimise costly operations Datrics caches the results of every brick during the pipeline run. After the pipeline is run - all the calculations are cached, so if the brick (arguments and inputs) remain the same brick will not be re-calculated again. This can speed-up the prototyping process since there’s no need to rerun again the entire pipeline.
However, it may slow down the pipeline running time. For big pipelines, you may disable caching for the pipeline. Try and see now much faster you’ll get to your result.
