Adding custom features to the Datrics pipeline and more
Introducing new cool features in the latest Datrics updates for data analysts to add custom bricks, benefit from an expanded toolset to analyze time series, and run experiments faster.
Enrich datrics.ai capabilities with custom features for your team
No code data analytics platforms, despite their simplification of access to data analysis, come together with their limitations. In particular, analysts are bound to use only methods, models, and data preparation instruments provided by the platform. Therefore users are forced to come up with workarounds, use multiple platforms, etc.
At Datrics, we aim to remove the functional restrictions with a new custom code brick editor. Our no-code analytics platform becomes more flexible and allows users to enhance the tool with new components.
Create custom code bricks to:
enhance the pipeline with new features
use new bricks in multiple pipelines within the project
share your custom bricks with the team
Custom code brick wraps users' Python code into the Datrics brick and uses it in the pipeline in a usual way. New brick follows the common brick structure and consists of:
Custom code brick can have multiple inputs and outputs of the data type. The code follows the Datrics bricks' structure with inputs and arguments validation methods, and brick execution method. An analyst may use the built-in functions to perform some inputs and arguments handling and metric calculation functions.
Just like any other brick within the no-code pipeline, custom code can be configured with arguments. Arguments are configured in the brick settings on the pipeline scene. This allows users to create more generic functionality in custom code and configure it specifically for the pipeline via the settings.
The fast calculation is critical for the data processing pipelines, especially in the case of a big data volume and the necessity to run the multiple repeated processes, for example, in case the problem of the imitation simulation class is solved. In the latest Datrics update, we drastically decrease the waiting time of the simple operations, with Math formula and Apply function bricks to run up to 5X faster.
New bricks to analyze data series
Iterative calculation brick
We added new brick that applies exponential weighted and cumulative functions to a selected column. Cumulative functions added: sum, min, max, product, and exponentially weighted functions: average, standard deviation, variance. Brick adds a new column to the dataset with the short name of the operation performed.
Analyzing time series data, we often need to approximate the values with the trendline. A new local approximation brick in our no-code analytics platform describes data with the selected trend line and window size. Brick performs approximation for a selected column with specified trend type returning approximated values. As a result, a dataset is updated with a new column.
Local trend analysis
Local trend analysis performs window-based approximation for a selected column with specified trend type returning coefficients of the approximation. Brick returns a dataset with coefficients of the approximation curve and an R2 score calculated for each point.
Teams with many projects and use cases on Datrics may schedule the pipeline runs or deploy an API to call it externally. It may become a challenge for data analytics platform admin to follow and manage the increasing list of deployed pipelines.Therefore, in datrics.ai we have added a place to manage deployed pipelines and links from all the projects and users. Admin may also deploy and undeploy pipelines owned by any user on the platform. The feature provides the means for the environment admin to react promptly to the issues.