Data visualization

CatBoost provides tools for the Python package that allow plotting charts with different training statistics. This information can be accessed both during and after the training procedure. Additional packages must be installed to support the visualization tools.

The following information is reflected on the charts:
  • metric values
  • best metric values on the test dataset
  • elapsed time of training
  • remaining time of training
  • current metric value
  • metric value on the best iteration
The table below lists the Python training parameters that affect visualization.
ParameterUsage tips
plotSet to “true”
name

The given value is used for signing the charts of the corresponding experiment. This parameter is useful when viewing results of different experiments on one chart.

custom_metric, loss_function, eval_metricAll the metrics specified in these parameters are output.
The following applications can be used for viewing the charts: