DJSW.evaluate

Functions

recall(y_true, y_pred[, average])	Calculate recall for multi-class classification.
precision(y_true, y_pred[, average])	Calculate precision for multi-class classification
accuracy(y_true, y_pred)	Calculate accuracy for multi-class classification
balanced_accuracy(y_true, y_pred)	Calculate balanced accuracy for multi-class classification.
f1_score(y_true, y_pred)	Calculate f1 for multi-class classification.
evaluate_model(args)	Evaluate a users model on their test data.
eval_model(args)

Module Contents

DJSW.evaluate.recall(y_true, y_pred, average='macro')

Calculate recall for multi-class classification.

Ratio of relevant retrieved instances to number of relevant instances.

Args:

y_true: Ground truth labels (1D array-like) y_pred: Predicted labels (1D array-like) average: Averaging mode, macro or micro (str)

Returns:

float: Recall score between 0 and 1

Raises:

AssertionError: If inputs are invalid ValueError: If computation fails

DJSW.evaluate.precision(y_true, y_pred, average='macro')

Calculate precision for multi-class classification

Ratio of relevant retrieved instances to number of retrieved instances

Args:

y_true: Ground truth labels (1D array-like) y_pred: Predicted labels (1D array-like) average: Averaging mode, macro or micro (str)

Returns:

float: Accuracy score between 0 and 1

Raises:

AssertionError: If inputs are invalid ValueError: If computation fails

DJSW.evaluate.accuracy(y_true, y_pred)

Calculate accuracy for multi-class classification

Ratio of correct classifications to number of classifications

Args:

y_true: Ground truth labels (1D array-like) y_pred: Predicted labels (1D array-like)

Returns:

float: Accuracy score between 0 and 1

Raises:

AssertionError: If inputs are invalid ValueError: If computation fails

DJSW.evaluate.balanced_accuracy(y_true, y_pred)

Calculate balanced accuracy for multi-class classification.

Balanced accuracy is the average of recall obtained on each class. It’s particularly useful for imbalanced datasets.

Args:

y_true: Ground truth labels (1D array-like) y_pred: Predicted labels (1D array-like)

Returns:

float: Balanced accuracy score between 0 and 1

Raises:

AssertionError: If inputs are invalid ValueError: If computation fails

DJSW.evaluate.f1_score(y_true, y_pred)

Calculate f1 for multi-class classification.

Combines precision and recall using the harmonic mean

Args:

y_true: Ground truth labels (1D array-like) y_pred: Predicted labels (1D array-like)

Returns:

float: F1 score between 0 and 1

Raises:

AssertionError: If inputs are invalid ValueError: If computation fails

DJSW.evaluate.evaluate_model(args)

Evaluate a users model on their test data.

DJSW.evaluate.eval_model(args)