A confusion matrix comparing predicted vs actual group labels. More...

#include <ConfusionMatrix.hpp>

Public Member Functions
	ConfusionMatrix ()=default
	Default-construct an empty confusion matrix.

	ConfusionMatrix (types::ResponseVector const &predictions, types::ResponseVector const &actual)
	Construct a confusion matrix from predictions and actual labels.

float	error () const
	Compute the overall error rate (1 - accuracy).

types::Vector< float >	group_errors () const
	Compute per-group error rates.

Public Attributes
std::map< int, int >	label_index
	Map from group label to matrix row/column index.

types::Matrix< int >	values
	The NxN confusion matrix (actual x predicted).

Detailed Description

A confusion matrix comparing predicted vs actual group labels.

Rows correspond to actual groups and columns to predicted groups. Provides overall error rate, per-group error rates, JSON serialization, and formatted terminal printing with diagonal highlighting and marginal errors.

types::ResponseVector preds = model.predict(x_test);
ConfusionMatrix cm(preds, y_test);
 
float err = cm.error();                  // overall error rate
auto per_class = cm.group_errors();      // per-group error rates

Constructor & Destructor Documentation

◆ ConfusionMatrix() [1/2]

ppforest2::stats::ConfusionMatrix::ConfusionMatrix ( )

default

Default-construct an empty confusion matrix.

◆ ConfusionMatrix() [2/2]

ppforest2::stats::ConfusionMatrix::ConfusionMatrix	(	types::ResponseVector const &	predictions,
		types::ResponseVector const &	actual )

Construct a confusion matrix from predictions and actual labels.

Parameters

predictions	The predicted group labels.
actual	The true group labels (must have the same size).

Exceptions

std::invalid_argument If predictions and actual have different sizes.

Member Function Documentation

◆ error()

float ppforest2::stats::ConfusionMatrix::error ( ) const

Compute the overall error rate (1 - accuracy).

Returns: The proportion of misclassified samples.

◆ group_errors()

types::Vector< float > ppforest2::stats::ConfusionMatrix::group_errors ( ) const

Compute per-group error rates.

Returns: A vector of error rates (one per group), where 0 = perfect, 1 = all wrong.

Member Data Documentation

◆ label_index

std::map<int, int> ppforest2::stats::ConfusionMatrix::label_index

Map from group label to matrix row/column index.

◆ values

types::Matrix<int> ppforest2::stats::ConfusionMatrix::values

The NxN confusion matrix (actual x predicted).

The documentation for this struct was generated from the following file:

core/src/stats/ConfusionMatrix.hpp

Public Member Functions

Public Attributes