Struct QualityProfile 

pub struct QualityProfile {
    pub name: String,
    pub description: String,
    pub expected_constant_columns: HashSet<String>,
    pub nullable_columns: HashSet<String>,
    pub max_null_ratio: f64,
    pub max_duplicate_ratio: f64,
    pub min_cardinality: usize,
    pub max_outlier_ratio: f64,
    pub max_duplicate_row_ratio: f64,
    pub penalize_unexpected_constants: bool,
    pub require_signature: bool,
}

Quality profile for customizing scoring rules per data type.

Different data types (doctest corpora, ML training sets, time series, etc.) have different expectations. For example:

• Doctest corpus: source and version columns are expected to be constant
• ML training: features should have high variance, labels can be categorical
• Time series: timestamps should be unique and sequential

Example

let profile = QualityProfile::doctest_corpus();
let score = profile.score_report(&report);

Fields

name: String

Profile name for display

description: String

Description of what this profile is for

expected_constant_columns: HashSet<String>

Columns that are expected to be constant (not penalized)

nullable_columns: HashSet<String>

Columns where high null ratio is acceptable

max_null_ratio: f64

Maximum acceptable null ratio (default: 0.1)

max_duplicate_ratio: f64

Maximum acceptable duplicate ratio (default: 0.5)

min_cardinality: usize

Minimum cardinality before flagging as low (default: 2)

max_outlier_ratio: f64

Maximum outlier ratio to report (default: 0.05)

max_duplicate_row_ratio: f64

Maximum duplicate row ratio (default: 0.01)

penalize_unexpected_constants: bool

Whether to penalize constant columns not in expected list

require_signature: bool

Whether this profile requires a signature column (for doctest)

Implementations

impl QualityProfile

pub fn new(name: impl Into<String>) -> Self

Create a new profile with custom name

pub fn by_name(name: &str) -> Option<Self>

Get profile by name

pub fn available_profiles() -> Vec<&'static str>

List available profile names

pub fn doctest_corpus() -> Self

Doctest corpus profile - for Python doctest extraction datasets.

Expects:

• source and version columns to be constant (single crate/version)
• signature column may have nulls (module-level doctests)
• input, expected, function should be non-null

pub fn ml_training() -> Self

ML training profile - for machine learning datasets.

Expects:

• Features to have reasonable variance
• Labels can be categorical (low cardinality OK)
• No null values in features or labels

pub fn time_series() -> Self

Time series profile - for temporal data.

Expects:

• Timestamp column should be unique
• Data should have temporal patterns

pub fn with_description(self, desc: impl Into<String>) -> Self

Set description

pub fn with_expected_constant(self, column: impl Into<String>) -> Self

Add an expected constant column

pub fn with_nullable(self, column: impl Into<String>) -> Self

Add a nullable column

pub fn with_max_null_ratio(self, ratio: f64) -> Self

Set max null ratio

pub fn with_max_duplicate_ratio(self, ratio: f64) -> Self

Set max duplicate ratio

pub fn is_expected_constant(&self, column: &str) -> bool

Check if a column is expected to be constant

pub fn is_nullable(&self, column: &str) -> bool

Check if a column is allowed to have nulls

pub fn null_threshold_for(&self, column: &str) -> f64

Get effective null threshold for a column

Trait Implementations

impl Clone for QualityProfile

fn clone(&self) -> QualityProfile

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for QualityProfile

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for QualityProfile

fn default() -> Self

Returns the “default value” for a type.