Transformation#

Data transformation utilities for Nsight Python profiling output.

This module contains functions that process raw profiling results, aggregate metrics, normalize them, and prepare the data for visualization or further statistical analysis.

nsight.transformation.aggregate_data(
df: DataFrame,
func: Callable[[...], Any],
normalize_against: str | None,
output_progress: bool,
)#

Groups and aggregates profiling data by configuration and annotation.

Parameters:

  • df (DataFrame) – The raw profiling results.

  • func (Callable[..., Any]) – Function representing kernel configuration parameters.

  • normalize_against (str | None) – Name of the annotation to normalize against.

  • output_progress (bool) – Toggles the display of data processing logs

Return type:

DataFrame

Returns:

Aggregated DataFrame and the (possibly normalized) metric name.