coonfit.helper#

Utility functions for the coonfit regression workflow.

This module provides supporting functions used during predictor validation and data quality assessment. It includes tools for detecting rank-deficient predictor matrices (which would cause the normal equations to be singular) and for counting the number of usable pixels within a boolean selector mask.

Functions#

check_rank_deficiency(array[, return_by_issue_type])

Check if matrix is rank deficient and identify problematic columns.

usable_pixels_count(selector)

Count the number of usable pixels determined by the selector.

usable_pixels_info(all_pixels, data_pixels)

Print the fraction of usable pixels.

Module Contents#

coonfit.helper.check_rank_deficiency(array, return_by_issue_type=False)[source]#

Check if matrix is rank deficient and identify problematic columns.

Returns a dictionary with column indices (key) and issue description (value). An empty dictionary indicates that no rank deficiency was detected.

Parameters:

  • array (NDArray) – Matrix to check for rank deficiency

  • return_by_issue_type (bool, optional) – If True, returns nested dictionary separating issues by type: {“linear_dependent”: […], “all_zero”: […]}

Returns:

Problematic columns and their issues. Uses numpy.linalg.matrix_rank() to determine the rank of the array.

Return type:

dict[int, str] or dict[str, list[int]]

See also

get_XT_X_dependency()

Check predictors for linear dependency.

coonfit.helper.usable_pixels_count(selector)[source]#

Count the number of usable pixels determined by the selector.

Parameters:

selector (NDArray) – Boolean array where True indicates a usable pixel and False indicates a pixel to be excluded

Returns:

Number of True values in the selector array (count of usable pixels). Uses numpy.unique() to count occurrences.

Return type:

int

See also

usable_pixels_info()

Print the fraction of usable pixels.

Examples

>>> selector = np.array([True, True, False, True, False])
>>> usable_pixels_count(selector)
3
coonfit.helper.usable_pixels_info(all_pixels, data_pixels)[source]#

Print the fraction of usable pixels.

Parameters:

  • all_pixels (int) – Total number of pixels in the dataset

  • data_pixels (int) – Number of pixels that contain usable data

Return type:

None

See also

usable_pixels_count()

Count the number of usable pixels.

Examples

>>> usable_pixels_info(1000, 750)
Of all_pixels=1000 there are data_pixels=750, i.e. 75.0% are usable