Create a custom report
TL;DR: You can create a custom report from 35+ individual metrics available in the library.
1. Choose metrics
To design a custom report, you should first define which metrics you want to include.
You can use Metric Presets as a starting point to explore types of analysis available in the library. Note that there are additional metrics that are not included in the presets that you can choose from.
Reference: The complete list of metrics is available in the All metrics table. To see interactive examples, refer to the Example notebooks.
Dataset-level metrics
Some of the metrics are calculated on the dataset level. For example, there is a metric that evaluates data drift for the whole dataset.
To create a custom report with dataset-level metrics, you need to create a Report
object and list the metrics
:
To calculate the metrics and get the visual report:
Column-level metrics
Some of the metrics are calculated on the column level. For example, to calculate drift in a specific feature or model predictions.
To create a custom report with column-level metrics:
To calculate the metrics and get the visual report:
Combining metrics and presets. When you define the composition of the Report, you can include metrics presets and individual tests in the same list. You can also combine feature-level and dataset-level metrics.
2. Set metric parameters
Some metrics include default parameters. For example, probabilistic classification quality metrics have a default 0.5 decision threshold. You can override the defaults by setting custom parameters.
Some metrics have required parameters. For example, if you want to calculate the number of values that match a regular expression, you need to specify it.
Example 1. How to specify regular expression (required parameter):
Example 2. How to specify a custom DataDrift test (optional parameter).
Reference: The available parameters for each metric are listed in the All metrics table.
Last updated