cursus.processing.validation

Field type validation utilities for preprocessing pipelines.

Provides strict validation functions that validate field types before applying numerical imputation or categorical risk table mapping.

validate_categorical_fields(df, cat_fields, dataset_name='dataset')

Strictly validate categorical fields before risk table mapping.

Parameters:

• df (DataFrame) – Input dataframe

• cat_fields (List[str]) – List of categorical field names from config

• dataset_name (str) – Name of dataset for error messages (e.g., “train”, “val”, “test”)

Raises:

• ValueError – If field not found in dataframe

• TypeError – If field has wrong type with specific field names

validate_numerical_fields(df, num_fields, dataset_name='dataset')

Strictly validate numerical fields before numerical imputation.

Parameters:

• df (DataFrame) – Input dataframe

• num_fields (List[str]) – List of numerical field names from config

• dataset_name (str) – Name of dataset for error messages (e.g., “train”, “val”, “test”)

Raises:

• ValueError – If field not found in dataframe

• TypeError – If field has wrong type with specific field names