count#
Methods to estimate latent structures used for confident learning, including:
Latent prior of the unobserved, errorless labels: py:
p(y)
Latent noisy channel (noise matrix) characterizing the flipping rates: nm:
P(given label  true label)
Latent inverse noise matrix characterizing the flipping process: inv:
P(true label  given label)
Latent confident_joint, an unnormalized matrix that counts the confident subset of label errors under the joint distribution for true/given label
These are estimated from a classification dataset. This module considers two types of datasets:
standard (multiclass) classification where each example is labeled as belonging to exactly one of K classes (e.g.
labels = np.array([0,0,1,0,2,1])
)multilabel classification where each example can be labeled as belonging to multiple classes (e.g.
labels = [[1,2],[1],[0],[],...]
)
Functions:

Calibrates any confident joint estimate 

Estimates the confident counts of latent true vs observed noisy labels for the examples in our dataset. 
Estimates 


This function computes the outofsample predicted probability [P(label=kx)] for every example in X using cross validation. 

Estimates the joint distribution of label noise 

Computes the latent prior 

Estimates the noise_matrix of shape 
Computes the confident counts estimate of latent variables py and the noise rates using observed labels and predicted probabilities, pred_probs. 

This function computes the outofsample predicted probability 


Returns expected (average) "selfconfidence" for each class. 

Estimates the number of label issues in a classification dataset. 
 cleanlab.count.calibrate_confident_joint(confident_joint, labels, *, multi_label=False)[source]#
Calibrates any confident joint estimate
P(label=i, true_label=j)
such thatnp.sum(cj) == len(labels)
andnp.sum(cj, axis = 1) == np.bincount(labels)
.In other words, this function forces the confident joint to have the true noisy prior
p(labels)
(summed over columns for each row) and also forces the confident joint to add up to the total number of examples.This method makes the confident joint a valid counts estimate of the actual joint of noisy and true labels.
 Parameters:
confident_joint (
np.ndarray
) – An array of shape(K, K)
representing the confident joint, the matrix used for identifying label issues, which estimates a confident subset of the joint distribution of the noisy and true labels,P_{noisy label, true label}
. Entry(j, k)
in the matrix is the number of examples confidently counted into the pair of(noisy label=j, true label=k)
classes. The confident_joint can be computed usingcount.compute_confident_joint
. If not provided, it is computed from the given (noisy) labels and pred_probs. If multi_label is True, then the confident_joint should be a onevsrest array of shape(K, 2, 2)
, and an array of the same shape will be returned.labels (
np.ndarray
orlist
) – Given class labels for each example in the dataset, some of which may be erroneous, in same format expected byfilter.find_label_issues
function.multi_label (
bool
, optional) – IfFalse
, dataset is for regular (multiclass) classification, where each example belongs to exactly one class. IfTrue
, dataset is for multilabel classification, where each example can belong to multiple classes. See documentation ofcompute_confident_joint
for details. In multilabel classification, the confident/calibrated joint arrays have shape(K, 2, 2)
formatted in a onevsrest fashion such that they contain a 2x2 matrix for each class that counts examples which are correctly/incorrectly labeled as belonging to that class. After calibration, the entries in each classspecific 2x2 matrix will sum to the number of examples.
 Return type:
ndarray
 Returns:
calibrated_cj (
np.ndarray
) – An array of shape(K, K)
representing a valid estimate of the joint counts of noisy and true labels (if multi_label is False). If multi_label is True, the returned calibrated_cj is instead an onevsrest array of shape(K, 2, 2)
, where for class c: entry(c, 0, 0)
in this onevsrest array is the number of examples whose noisy label contains c confidently identified as truly belonging to class c as well. Entry(c, 1, 0)
in this onevsrest array is the number of examples whose noisy label contains c confidently identified as not actually belonging to class c. Entry(c, 0, 1)
in this onevsrest array is the number of examples whose noisy label does not contain c confidently identified as truly belonging to class c. Entry(c, 1, 1)
in this onevsrest array is the number of examples whose noisy label does not contain c confidently identified as actually not belonging to class c as well.
 cleanlab.count.compute_confident_joint(labels, pred_probs, *, thresholds=None, calibrate=True, multi_label=False, return_indices_of_off_diagonals=False)[source]#
Estimates the confident counts of latent true vs observed noisy labels for the examples in our dataset. This array of shape
(K, K)
is called the confident joint and contains counts of examples in every class, confidently labeled as every other class. These counts may subsequently be used to estimate the joint distribution of true and noisy labels (by normalizing them to frequencies).Important: this function assumes that pred_probs are outofsample holdout probabilities. This can be done with cross validation. If the probabilities are not computed outofsample, overfitting may occur.
 Parameters:
labels (
np.ndarray
orlist
) – Given class labels for each example in the dataset, some of which may be erroneous, in same format expected byfilter.find_label_issues
function.pred_probs (
np.ndarray
) – Modelpredicted class probabilities for each example in the dataset, in same format expected byfilter.find_label_issues
function.thresholds (
array_like
, optional) –An array of shape
(K, 1)
or(K,)
of perclass threshold probabilities, used to determine the cutoff probability necessary to consider an example as a given class label (see Northcutt et al., 2021, Section 3.1, Equation 2).This is for advanced users only. If not specified, these are computed for you automatically. If an example has a predicted probability greater than this threshold, it is counted as having true_label = k. This is not used for pruning/filtering, only for estimating the noise rates using confident counts.
calibrate (
bool
, defaultTrue
) – Calibrates confident joint estimateP(label=i, true_label=j)
such thatnp.sum(cj) == len(labels)
andnp.sum(cj, axis = 1) == np.bincount(labels)
. Whencalibrate=True
, this method returns an estimate of the latent true joint counts of noisy and true labels.multi_label (
bool
, optional) – IfTrue
, this is multilabel classification dataset (where each example can belong to more than one class) rather than a regular (multiclass) classifiction dataset. In this case, labels should be an iterable (e.g. list) of iterables (e.g.List[List[int]]
), containing the list of classes to which each example belongs, instead of just a single class. Example of labels for a multilabel classification dataset:[[0,1], [1], [0,2], [0,1,2], [0], [1], [], ...]
.return_indices_of_off_diagonals (
bool
, optional) – IfTrue
, returns indices of examples that were counted in offdiagonals of confident joint as a baseline proxy for the label issues. This sometimes works as well asfilter.find_label_issues(confident_joint)
.
 Return type:
Union
[ndarray
,Tuple
[ndarray
,list
]] Returns:
 confident_joint_counts (
np.ndarray
) – An array of shape(K, K)
representing counts of examples for which we are confident about their given and true label (if multi_label is False). If multi_label is True, this array instead has shape
(K, 2, 2)
representing a onevsrest format for the confident joint, where for each class c: Entry(c, 0, 0)
in this onevsrest array is the number of examples whose noisy label contains c confidently identified as truly belonging to class c as well. Entry(c, 1, 0)
in this onevsrest array is the number of examples whose noisy label contains c confidently identified as not actually belonging to class c. Entry(c, 0, 1)
in this onevsrest array is the number of examples whose noisy label does not contain c confidently identified as truly belonging to class c. Entry(c, 1, 1)
in this onevsrest array is the number of examples whose noisy label does not contain c confidently identified as actually not belonging to class c as well.
If return_indices_of_off_diagonals is set as True, this function instead returns a tuple (confident_joint, indices_off_diagonal) where indices_off_diagonal is a list of arrays and each array contains the indices of examples counted in offdiagonals of confident joint.
 confident_joint_counts (
Note
We provide a forloop based simplification of the confident joint below. This implementation is not efficient, not used in practice, and not complete, but covers the gist of how the confident joint is computed:
# Confident examples are those that we are confident have true_label = k # Estimate (K, K) matrix of confident examples with label = k_s and true_label = k_y cj_ish = np.zeros((K, K)) for k_s in range(K): # k_s is the class value k of noisy labels `s` for k_y in range(K): # k_y is the (guessed) class k of true_label k_y cj_ish[k_s][k_y] = sum((pred_probs[:,k_y] >= (thresholds[k_y]  1e8)) & (labels == k_s))
The following is a vectorized (but nonparallelized) implementation of the confident joint, again slow, using forloops/simplified for understanding. This implementation is 100% accurate, it’s just not optimized for speed.
confident_joint = np.zeros((K, K), dtype = int) for i, row in enumerate(pred_probs): s_label = labels[i] confident_bins = row >= thresholds  1e6 num_confident_bins = sum(confident_bins) if num_confident_bins == 1: confident_joint[s_label][np.argmax(confident_bins)] += 1 elif num_confident_bins > 1: confident_joint[s_label][np.argmax(row)] += 1
 cleanlab.count.estimate_confident_joint_and_cv_pred_proba(X, labels, clf=LogisticRegression(), *, cv_n_folds=5, thresholds=None, seed=None, calibrate=True, clf_kwargs={}, validation_func=None)[source]#
Estimates
P(labels, y)
, the confident counts of the latent joint distribution of true and noisy labels using observed labels and predicted probabilities pred_probs.The output of this function is an array of shape
(K, K)
.Under certain conditions, estimates are exact, and in many conditions, estimates are within one percent of actual.
Notes: There are two ways to compute the confident joint with pros/cons. (1) For each holdout set, we compute the confident joint, then sum them up. (2) Compute pred_proba for each fold, combine, compute the confident joint. (1) is more accurate because it correctly computes thresholds for each fold (2) is more accurate when you have only a little data because it computes the confident joint using all the probabilities. For example if you had 100 examples, with 5fold cross validation + uniform p(y) you would only have 20 examples to compute each confident joint for (1). Such small amounts of data is bound to result in estimation errors. For this reason, we implement (2), but we implement (1) as a commented out function at the end of this file.
 Parameters:
X (
np.ndarray
orpd.DataFrame
) –Input feature matrix of shape
(N, ...)
, where N is the number of examples. The classifier that this instance was initialized with,clf
, must be able to fit() and predict() data with this format.labels (
np.ndarray
orpd.Series
) – A 1D array of shape(N,)
containing class labels for a standard (multiclass) classification dataset. Some given labels may be erroneous. Elements must be integers in the set 0, 1, …, K1, where K is the number of classes. All classes must be present in the dataset.clf (
estimator instance
, optional) – A classifier implementing the sklearn estimator API.cv_n_folds (
int
, default5
) – The number of crossvalidation folds used to compute outofsample predicted probabilities for each example in X.thresholds (
array_like
, optional) –An array of shape
(K, 1)
or(K,)
of perclass threshold probabilities, used to determine the cutoff probability necessary to consider an example as a given class label (see Northcutt et al., 2021, Section 3.1, Equation 2).This is for advanced users only. If not specified, these are computed for you automatically. If an example has a predicted probability greater than this threshold, it is counted as having true_label = k. This is not used for pruning/filtering, only for estimating the noise rates using confident counts.
seed (
int
, optional) – Set the default state of the random number generator used to split the crossvalidated folds. If None, uses np.random current random state.calibrate (
bool
, defaultTrue
) – Calibrates confident joint estimateP(label=i, true_label=j)
such thatnp.sum(cj) == len(labels)
andnp.sum(cj, axis = 1) == np.bincount(labels)
.clf_kwargs (
dict
, optional) – Optional keyword arguments to pass into clf’sfit()
method.validation_func (
callable
, optional) – Specifies how to map the validation data split in crossvalidation as input forclf.fit()
. For details, see the documentation ofCleanLearning.fit
 Return type:
Tuple
[ndarray
,ndarray
] Returns:
estimates (
tuple
) – Tuple of two numpy arrays in the form: (joint counts matrix, predicted probability matrix)
Note
Multilabel classification is not supported in this method.
 cleanlab.count.estimate_cv_predicted_probabilities(X, labels, clf=LogisticRegression(), *, cv_n_folds=5, seed=None, clf_kwargs={}, validation_func=None)[source]#
This function computes the outofsample predicted probability [P(label=kx)] for every example in X using cross validation. Output is a np.ndarray of shape
(N, K)
where N is the number of training examples and K is the number of classes. Parameters:
X (
np.ndarray
) – Input feature matrix of shape(N, ...)
, where N is the number of examples. The classifier that this instance was initialized with, clf, must be able to handle data with this shape.labels (
np.ndarray
) – A 1D array of shape(N,)
containing class labels for a standard (multiclass) classification dataset. Some given labels may be erroneous. Elements must be integers in the set 0, 1, …, K1, where K is the number of classes. All classes must be present in the dataset.clf (
estimator instance
, optional) –A classifier implementing the sklearn estimator API.
cv_n_folds (
int
, default5
) – The number of crossvalidation folds used to compute outofsample probabilities for each example in X.seed (
int
, optional) – Set the default state of the random number generator used to split the crossvalidated folds. IfNone
, usesnp.random
current random state.clf_kwargs (
dict
, optional) – Optional keyword arguments to pass into clf’sfit()
method.validation_func (
callable
, optional) – Specifies how to map the validation data split in crossvalidation as input forclf.fit()
. For details, see the documentation ofCleanLearning.fit
 Return type:
ndarray
 Returns:
pred_probs (
np.ndarray
) – An array of shape(N, K)
representingP(label=kx)
, the modelpredicted probabilities. Each row of this matrix corresponds to an example x and contains the modelpredicted probabilities that x belongs to each possible class.
 cleanlab.count.estimate_joint(labels, pred_probs, *, confident_joint=None, multi_label=False)[source]#
Estimates the joint distribution of label noise
P(label=i, true_label=j)
guaranteed to:Sum to 1
Satisfy
np.sum(joint_estimate, axis = 1) == p(labels)
 Parameters:
labels (
np.ndarray
orlist
) – Given class labels for each example in the dataset, some of which may be erroneous, in same format expected byfilter.find_label_issues
function.pred_probs (
np.ndarray
) – Modelpredicted class probabilities for each example in the dataset, in same format expected byfilter.find_label_issues
function.confident_joint (
np.ndarray
, optional) – Array of estimated class label error statisics used for identifying label issues, in same format expected byfilter.find_label_issues
function. The confident_joint can be computed usingcount.compute_confident_joint
. If not provided, it is internally computed from the given (noisy) labels and pred_probs.multi_label (
bool
, optional) – IfFalse
, dataset is for regular (multiclass) classification, where each example belongs to exactly one class. IfTrue
, dataset is for multilabel classification, where each example can belong to multiple classes. See documentation ofcompute_confident_joint
for details.
 Return type:
ndarray
 Returns:
confident_joint_distribution (
np.ndarray
) – An array of shape(K, K)
representing an estimate of the true joint distribution of noisy and true labels (if multi_label is False). If multi_label is True, an array of shape(K, 2, 2)
representing an estimate of the true joint distribution of noisy and true labels for each class in a onevsrest fashion. Entry(c, i, j)
in this array is the number of examples confidently counted into a(class c, noisy label=i, true label=j)
bin, where i, j are either 0 or 1 to denote whether this example belongs to class c or not (recall examples can belong to multiple classes in multilabel classification).
 cleanlab.count.estimate_latent(confident_joint, labels, *, py_method='cnt', converge_latent_estimates=False)[source]#
Computes the latent prior
p(y)
, the noise matrixP(labelsy)
and the inverse noise matrixP(ylabels)
from the confident_jointcount(labels, y)
. The confident_joint can be estimated by compute_confident_joint <cleanlab.count.compute_confident_joint> which counts confident examples. Parameters:
confident_joint (
np.ndarray
) – An array of shape(K, K)
representing the confident joint, the matrix used for identifying label issues, which estimates a confident subset of the joint distribution of the noisy and true labels,P_{noisy label, true label}
. Entry(j, k)
in the matrix is the number of examples confidently counted into the pair of(noisy label=j, true label=k)
classes. The confident_joint can be computed usingcount.compute_confident_joint
. If not provided, it is computed from the given (noisy) labels and pred_probs.labels (
np.ndarray
) – A 1D array of shape(N,)
containing class labels for a standard (multiclass) classification dataset. Some given labels may be erroneous. Elements must be integers in the set 0, 1, …, K1, where K is the number of classes.py_method (
{"cnt", "eqn", "marginal", "marginal_ps"}
, default"cnt"
) – py is shorthand for the “class proportions (a.k.a prior) of the true labels”. This method defines how to compute the latent priorp(true_label=k)
. Default is"cnt"
, which works well even when the noise matrices are estimated poorly by using the matrix diagonals instead of all the probabilities.converge_latent_estimates (
bool
, optional) – IfTrue
, forces numerical consistency of estimates. Each is estimated independently, but they are related mathematically with closed form equivalences. This will iteratively make them mathematically consistent.
 Return type:
Tuple
[ndarray
,ndarray
,ndarray
] Returns:
tuple
– A tuple containing (py, noise_matrix, inv_noise_matrix).
Note
Multilabel classification is not supported in this method.
 cleanlab.count.estimate_noise_matrices(X, labels, clf=LogisticRegression(), *, cv_n_folds=5, thresholds=None, converge_latent_estimates=True, seed=None, clf_kwargs={}, validation_func=None)[source]#
Estimates the noise_matrix of shape
(K, K)
. This is the fraction of examples in every class, labeled as every other class. The noise_matrix is a conditional probability matrix forP(label=k_strue_label=k_y)
.Under certain conditions, estimates are exact, and in most conditions, estimates are within one percent of the actual noise rates.
 Parameters:
X (
np.ndarray
) – Input feature matrix of shape(N, ...)
, where N is the number of examples. The classifier that this instance was initialized with, clf, must be able to handle data with this shape.labels (
np.ndarray
) – An array of shape(N,)
of noisy labels, i.e. some labels may be erroneous. Elements must be integers in the set 0, 1, …, K1, where K is the number of classes.clf (
estimator instance
, optional) –A classifier implementing the sklearn estimator API.
cv_n_folds (
int
, default5
) – The number of crossvalidation folds used to compute outofsample probabilities for each example in X.thresholds (
array_like
, optional) –An array of shape
(K, 1)
or(K,)
of perclass threshold probabilities, used to determine the cutoff probability necessary to consider an example as a given class label (see Northcutt et al., 2021, Section 3.1, Equation 2).This is for advanced users only. If not specified, these are computed for you automatically. If an example has a predicted probability greater than this threshold, it is counted as having true_label = k. This is not used for pruning/filtering, only for estimating the noise rates using confident counts.
converge_latent_estimates (
bool
, optional) – IfTrue
, forces numerical consistency of estimates. Each is estimated independently, but they are related mathematically with closed form equivalences. This will iteratively make them mathematically consistent.seed (
int
, optional) – Set the default state of the random number generator used to split the crossvalidated folds. If None, uses np.random current random state.clf_kwargs (
dict
, optional) – Optional keyword arguments to pass into clf’sfit()
method.validation_func (
callable
, optional) – Specifies how to map the validation data split in crossvalidation as input forclf.fit()
. For details, see the documentation ofCleanLearning.fit
 Return type:
Tuple
[ndarray
,ndarray
] Returns:
estimates (
tuple
) – A tuple containing arrays (noise_matrix, inv_noise_matrix).
 cleanlab.count.estimate_py_and_noise_matrices_from_probabilities(labels, pred_probs, *, thresholds=None, converge_latent_estimates=True, py_method='cnt', calibrate=True)[source]#
Computes the confident counts estimate of latent variables py and the noise rates using observed labels and predicted probabilities, pred_probs.
Important: this function assumes that pred_probs are outofsample holdout probabilities. This can be done with cross validation. If the probabilities are not computed outofsample, overfitting may occur.
This function estimates the noise_matrix of shape
(K, K)
. This is the fraction of examples in every class, labeled as every other class. The noise_matrix is a conditional probability matrix forP(label=k_strue_label=k_y)
.Under certain conditions, estimates are exact, and in most conditions, estimates are within one percent of the actual noise rates.
 Parameters:
labels (
np.ndarray
) – A 1D array of shape(N,)
containing class labels for a standard (multiclass) classification dataset. Some given labels may be erroneous. Elements must be integers in the set 0, 1, …, K1, where K is the number of classes.pred_probs (
np.ndarray
) – Modelpredicted class probabilities for each example in the dataset, in same format expected byfilter.find_label_issues
function.thresholds (
array_like
, optional) –An array of shape
(K, 1)
or(K,)
of perclass threshold probabilities, used to determine the cutoff probability necessary to consider an example as a given class label (see Northcutt et al., 2021, Section 3.1, Equation 2).This is for advanced users only. If not specified, these are computed for you automatically. If an example has a predicted probability greater than this threshold, it is counted as having true_label = k. This is not used for pruning/filtering, only for estimating the noise rates using confident counts.
converge_latent_estimates (
bool
, optional) – IfTrue
, forces numerical consistency of estimates. Each is estimated independently, but they are related mathematically with closed form equivalences. This will iteratively make them mathematically consistent.py_method (
{"cnt", "eqn", "marginal", "marginal_ps"}
, default"cnt"
) – How to compute the latent priorp(true_label=k)
. Default is"cnt"
as it often works well even when the noise matrices are estimated poorly by using the matrix diagonals instead of all the probabilities.calibrate (
bool
, defaultTrue
) – Calibrates confident joint estimateP(label=i, true_label=j)
such thatnp.sum(cj) == len(labels)
andnp.sum(cj, axis = 1) == np.bincount(labels)
.
 Return type:
Tuple
[ndarray
,ndarray
,ndarray
,ndarray
] Returns:
estimates (
tuple
) – A tuple of arrays: (py, noise_matrix, inverse_noise_matrix, confident_joint).
Note
Multilabel classification is not supported in this method.
 cleanlab.count.estimate_py_noise_matrices_and_cv_pred_proba(X, labels, clf=LogisticRegression(), *, cv_n_folds=5, thresholds=None, converge_latent_estimates=False, py_method='cnt', seed=None, clf_kwargs={}, validation_func=None)[source]#
This function computes the outofsample predicted probability
P(label=kx)
for every example x in X using cross validation while also computing the confident counts noise rates within each crossvalidated subset and returning the average noise rate across all examples.This function estimates the noise_matrix of shape
(K, K)
. This is the fraction of examples in every class, labeled as every other class. The noise_matrix is a conditional probability matrix forP(label=k_strue_label=k_y)
.Under certain conditions, estimates are exact, and in most conditions, estimates are within one percent of the actual noise rates.
 Parameters:
X (
np.ndarray
) – Input feature matrix of shape(N, ...)
, where N is the number of examples. The classifier that this instance was initialized with, clf, must be able to handle data with this shape.labels (
np.ndarray
) – A 1D array of shape(N,)
containing class labels for a standard (multiclass) classification dataset. Some given labels may be erroneous. Elements must be integers in the set 0, 1, …, K1, where K is the number of classes. All classes must be present in the dataset.clf (
estimator instance
, optional) –A classifier implementing the sklearn estimator API.
cv_n_folds (
int
, default5
) – The number of crossvalidation folds used to compute outofsample probabilities for each example in X.thresholds (
array_like
, optional) –An array of shape
(K, 1)
or(K,)
of perclass threshold probabilities, used to determine the cutoff probability necessary to consider an example as a given class label (see Northcutt et al., 2021, Section 3.1, Equation 2).This is for advanced users only. If not specified, these are computed for you automatically. If an example has a predicted probability greater than this threshold, it is counted as having true_label = k. This is not used for pruning/filtering, only for estimating the noise rates using confident counts.
converge_latent_estimates (
bool
, optional) – IfTrue
, forces numerical consistency of estimates. Each is estimated independently, but they are related mathematically with closed form equivalences. This will iteratively make them mathematically consistent.py_method (
{"cnt", "eqn", "marginal", "marginal_ps"}
, default"cnt"
) – How to compute the latent priorp(true_label=k)
. Default is"cnt"
as it often works well even when the noise matrices are estimated poorly by using the matrix diagonals instead of all the probabilities.seed (
int
, optional) – Set the default state of the random number generator used to split the crossvalidated folds. IfNone
, usesnp.random
current random state.clf_kwargs (
dict
, optional) – Optional keyword arguments to pass into clf’sfit()
method.validation_func (
callable
, optional) – Specifies how to map the validation data split in crossvalidation as input forclf.fit()
. For details, see the documentation ofCleanLearning.fit
 Return type:
Tuple
[ndarray
,ndarray
,ndarray
,ndarray
,ndarray
] Returns:
estimates (
tuple
) – A tuple of five arrays (py, noise matrix, inverse noise matrix, confident joint, predicted probability matrix).
Note
Multilabel classification is not supported in this method.
 cleanlab.count.get_confident_thresholds(labels, pred_probs, multi_label=False)[source]#
Returns expected (average) “selfconfidence” for each class.
The confident class threshold for a class j is the expected (average) “selfconfidence” for class j, i.e. the modelpredicted probability of this class averaged amongst all examples labeled as class j.
 Parameters:
labels (
np.ndarray
orlist
) – Given class labels for each example in the dataset, some of which may be erroneous, in same format expected byfilter.find_label_issues
function.pred_probs (
np.ndarray
) – Modelpredicted class probabilities for each example in the dataset, in same format expected byfilter.find_label_issues
function.multi_label (
bool
, default= False
) – SetFalse
if your dataset is for regular (multiclass) classification, where each example belongs to exactly one class. SetTrue
if your dataset is for multilabel classification, where each example can belong to multiple classes. See documentation ofcompute_confident_joint
for details.
 Return type:
ndarray
 Returns:
confident_thresholds (
np.ndarray
) – An array of shape(K, )
where K is the number of classes.
 cleanlab.count.num_label_issues(labels, pred_probs, *, confident_joint=None, estimation_method='off_diagonal', multi_label=False)[source]#
Estimates the number of label issues in a classification dataset. Use this method to get the most accurate estimate of number of label issues when you don’t need the indices of the examples with label issues.
 Parameters:
labels (
np.ndarray
orlist
) – Given class labels for each example in the dataset, some of which may be erroneous, in same format expected byfilter.find_label_issues
function.pred_probs (
ndarray
) – Modelpredicted class probabilities for each example in the dataset, in same format expected byfilter.find_label_issues
function.confident_joint (
Optional
[ndarray
]) – Array of estimated class label error statisics used for identifying label issues, in same format expected byfilter.find_label_issues
function. The confident_joint can be computed usingcount.compute_confident_joint
. If not provided, it is internally computed from the given (noisy) labels and pred_probs.estimation_method (
str
) – Method for estimating the number of label issues in dataset by counting the examples in the offdiagonal of the confident_joint
P(label=i, true_label=j)
. 'off_diagonal'
: Counts the number of examples in the offdiagonal of the confident_joint. Returns the same value assum(find_label_issues(filter_by='confident_learning'))
'off_diagonal_calibrated'
: Calibrates confident joint estimateP(label=i, true_label=j)
such that
np.sum(cj) == len(labels)
andnp.sum(cj, axis = 1) == np.bincount(labels)
before counting the number of examples in the offdiagonal. Number will always be equal to or greater thanestimate_issues='off_diagonal'
. You can use this value as the cutoff threshold used with ranking/scoring functions fromcleanlab.rank
with num_label_issues overestimation_method='off_diagonal'
in two cases:As we add more label and data quality scoring functions in
cleanlab.rank
, this approach will always work.If you have a custom score to rank your data by label quality and you just need to know the cutoff of likely label issues.
TL;DR: use this method to get the most accurate estimate of number of label issues when you don’t need the indices of the label issues.
 Method for estimating the number of label issues in dataset by counting the examples in the offdiagonal of the confident_joint
multi_label (
bool
, optional) – SetFalse
if your dataset is for regular (multiclass) classification, where each example belongs to exactly one class. SetTrue
if your dataset is for multilabel classification, where each example can belong to multiple classes. See documentation ofcompute_confident_joint
for details.
 Return type:
int
 Returns:
num_issues
– The estimated number of examples with label issues in the dataset.