我正试图在训练练习中实现一个有序分类器，并且遇到了一些麻烦。我不能使用一个对所有的分类器，因为我的类是有序的。没有序数分类器的功能，所以我在互联网上找到了下面的代码。(来源:https://towardsdatascience.com/simple技巧- -火车一个序数回归——-任何分类器- 6911183 - d2a3c)。
我对我应该如何使用它感到困惑…我有一个训练和测试数据集……但我怎么把它们结合起来呢?例如，对于逻辑回归，我知道你会有这样的代码:
model = LogisticRegression()
model.fit(x_train, y_train)
但是我如何使用这些代码呢?我怎么得到概率呢?
代码来自网站:
from sklearn.base import clone
class OrdinalClassifier():
def __init__(self, clf):
self.clf = clf
self.clfs = {}
def fit(self, X, y):
self.unique_class = np.sort(np.unique(y))
if self.unique_class.shape[0] > 2:
for i in range(self.unique_class.shape[0]-1):
# for each k - 1 ordinal value we fit a binary classification problem
binary_y = (y > self.unique_class[i]).astype(np.uint8)
clf = clone(self.clf)
clf.fit(X, binary_y)
self.clfs[i] = clf
def predict_proba(self, X):
clfs_predict = {k:self.clfs[k].predict_proba(X) for k in self.clfs}
predicted = []
for i,y in enumerate(self.unique_class):
if i == 0:
# V1 = 1 - Pr(y > V1)
predicted.append(1 - clfs_predict[y][:,1])
elif y in clfs_predict:
# Vi = Pr(y > Vi-1) - Pr(y > Vi)
predicted.append(clfs_predict[y-1][:,1] - clfs_predict[y][:,1])
else:
# Vk = Pr(y > Vk-1)
predicted.append(clfs_predict[y-1][:,1])
return np.vstack(predicted).T
def predict(self, X):
return np.argmax(self.predict_proba(X), axis=1)
在运行代码时遇到了一些错误，所以我对代码做了一些更改:
from sklearn.base import clone
import numpy as np
# Source:
# 1. https://stackoverflow.com/questions/66486947/how-to-use-ordinal-classifier
# 2. https://towardsdatascience.com/simple-trick-to-train-an-ordinal-regression-with-any-classifier-6911183d2a3c
class OrdinalClassifier():
def __init__(self, clf):
self.clf = clf
self.clfs = {}
def fit(self, X, y):
self.unique_class = np.sort(np.unique(y))
if self.unique_class.shape[0] > 2:
for i in range(self.unique_class.shape[0] - 1):
# for each k - 1 ordinal value we fit a binary classification problem
binary_y = (y > self.unique_class[i]).astype(np.uint8)
clf = clone(self.clf)
clf.fit(X, binary_y)
self.clfs[i] = clf
def predict_proba(self, X):
clfs_predict = {k: v.predict_proba(X) for k, v in self.clfs.items()}
predicted = []
for i, y in enumerate(self.unique_class):
if i == 0:
# V1 = 1 - Pr(y > V1)
predicted.append(1 - clfs_predict[i][:, 1])
elif y in clfs_predict:
# Vi = Pr(y > Vi-1) - Pr(y > Vi)
predicted.append(clfs_predict[i - 1][:, 1] - clfs_predict[i][:, 1])
else:
# Vk = Pr(y > Vk-1)
predicted.append(clfs_predict[i - 1][:, 1])
return np.vstack(predicted).T
def predict(self, X):
return self.unique_class[np.argmax(self.predict_proba(X), axis=1)]
回到你的问题:
我有一个训练和测试数据集…但我怎么把它们结合起来呢?
您可以轻松地实现如下代码:
knn = KNeighborsClassifier()
oc = OrdinalClassifier(knn)
oc.fit(X_train, y_train)
oc.predict(X_test)
输出将是测试集的预测类标签。因此，您可以调用sklearn的混淆矩阵来检查准确性等。
我怎么得到概率?
你可以得到每个类的概率如下:
oc.predict_proba(X_test)
你将得到numpy的二维数组中每个类的概率mxn维，其中m为实例数，n为类数
如果您查看文章的评论，您将看到所提出的算法没有正确地格式化，正如@Arindam Paul所解释的那样:"因为您正在从集成中的不同分类器中减去概率，因此有可能使单个概率变为负值。我对我的问题进行了测试，发现某些情况下的概率为负。
这里有一个解决方案@CloudDude: https://github.com/leeprevost/OrdinalClassifier
摘自:https://github.com/leeprevost/OrdinalClassifier
import numpy as np
import scipy.sparse as sp
from sklearn.base import BaseEstimator, ClassifierMixin, clone, is_classifier
from sklearn.base import MultiOutputMixin
from sklearn.base import MetaEstimatorMixin, is_regressor
from sklearn.utils.deprecation import deprecated
from sklearn.utils._tags import _safe_tags
from sklearn.utils.validation import _num_samples
from sklearn.utils.validation import check_is_fitted
from sklearn.utils.multiclass import (
_check_partial_fit_first_call,
type_of_target
)
from sklearn.utils.metaestimators import _safe_split, available_if
from sklearn.utils.fixes import delayed
from sklearn.multiclass import (
_fit_binary,
_fit_ovo_binary,
_estimators_has
)
from joblib import Parallel
_fit_ovr_binary = _fit_binary
from typing import Iterable
class OrdinalClassifier(
MultiOutputMixin, ClassifierMixin, MetaEstimatorMixin, BaseEstimator
):
"""Ordinal multiclass strategy.
This classifier is based on a "Simple Approach to Oridinal Classification"
by Frank and Hall as oultined in this paper.
https://www.cs.waikato.ac.nz/~eibe/pubs/ordinal_tech_report.pdf
Adapted Abstract:
Machine learning methods for classification problems commonly assume
that the class values are unordered. However, in many practical applications
the class values do exhibit a natural order—for example, when learning how to grade
or when classifying sentiment (disagree < neutral < agree), temperatures (cold <
warm < hot). The standard approach to ordinal classification converts the class
value into a numeric quantity and applies a regression learner to the transformed
data, translating the output back into a discrete class value in a post-processing
step. A disadvantage of this method is that it can only be applied in conjunction with a
regression scheme.
The method enables standard classification algorithms to make use of ordering information
in class attributes. The authors have shown in their work this classifier
outperforms the naive state.
The method utilizes a 'simple trick' to allow the underlying classifiers to take
advantage of the ordinal class information. First, the data is tranformed from a k-class
ordinal problem to a n_classes - 1 binary class problem. Training starts by deriving new datasets from
the original dataset, one for each of the n_classes -1 binary class attributes.
--------
Ordinal attribute A* with ordered values V1, V2, ..., Vk into n_classes-1 binary attrbutes,
one for each of the original attribute's first n_classes-1 values. The ith binary attribute
represents the test A* > Vi.
--------
@todo: should this stay in? My starting point was to use OvR as basis. (not tested)
OrdinalClassifier can also be used for multilabel classification. To use
this feature, provide an indicator matrix for the target `y` when calling
`.fit`. In other words, the target labels should be formatted as a 2D
binary (0/1) matrix, where [i, j] == 1 indicates the presence of label j
in sample i. This estimator uses the binary relevance method to perform
multilabel classification, which involves training one binary classifier
independently for each label.
Read more in the :ref:`User Guide <ovr_classification>`.
Parameters
----------
estimator : estimator object
An estimator object implementing :term:`fit` and one of
:term:`decision_function` or :term:`predict_proba`.
n_jobs : int, default=None
The number of jobs to use for the computation: the `n_classes`
k-1 (n-1) ordinal problems problems are computed in parallel.
``None`` means 1 unless in a :obj:`joblib.parallel_backend` context.
``-1`` means using all processors. See :term:`Glossary <n_jobs>`
for more details.
.. versionchanged:: v0.20
`n_jobs` default changed from 1 to None
reverse_classes : reorders classes to shift the importance of the classes (eg. hot>mild>cold)
class_order: override the default sorted(np.unique(y)) classes_ attribute to shift order of computation of ordinals.
@todo: add validation of class order.
Attributes (based on OvR classifier -- @todo: edit)
----------
estimators_ : list of `n_classes` - 1 estimators
Estimators used for predictions. Each classifies a derived y as produced by private method.
coef_ : ndarray of shape (1, n_features) or (n_classes, n_features)
Coefficient of the features in the decision function. This attribute
exists only if the ``estimators_`` defines ``coef_``.
.. deprecated:: 0.24
This attribute is deprecated in 0.24 and will
be removed in 1.1 (renaming of 0.26). If you use this attribute
in :class:`~sklearn.feature_selection.RFE` or
:class:`~sklearn.feature_selection.SelectFromModel`,
you may pass a callable to the `importance_getter`
parameter that extracts feature the importances
from `estimators_`.
intercept_ : ndarray of shape (1, 1) or (n_classes, 1)
If ``y`` is binary, the shape is ``(1, 1)`` else ``(n_classes, 1)``
This attribute exists only if the ``estimators_`` defines
``intercept_``.
.. deprecated:: 0.24
This attribute is deprecated in 0.24 and will
be removed in 1.1 (renaming of 0.26). If you use this attribute
in :class:`~sklearn.feature_selection.RFE` or
:class:`~sklearn.feature_selection.SelectFromModel`,
you may pass a callable to the `importance_getter`
parameter that extracts feature the importances
from `estimators_`.
classes_ : array, shape = [`n_classes`]
Class labels.
n_classes_ : int
Number of classes.
multilabel_ : boolean @todo: need to turn this off for now. Untested.
Whether a OrdinalClassifier is a multilabel classifier.
n_features_in_ : int
Number of features seen during :term:`fit`. Only defined if the
underlying estimator exposes such an attribute when fit.
.. versionadded:: 0.24
feature_names_in_ : ndarray of shape (`n_features_in_`,)
Names of features seen during :term:`fit`. Only defined if the
underlying estimator exposes such an attribute when fit.
.. versionadded:: 1.0
ordinal_prob_names_ : generated by predict_proba method. List of ordinal probability
names that correspond to the paper an to the order of classes_ of n_classes length
See Also
--------
MultiOutputClassifier : Alternate way of extending an estimator for
multilabel classification.
Examples (@todo: redo)
--------
>>> import numpy as np
>>> from sklearn.multiclass import OneVsRestClassifier
>>> from sklearn.svm import SVC
>>> X = np.array([
... [10, 10],
... [8, 10],
... [-5, 5.5],
... [-5.4, 5.5],
... [-20, -20],
... [-15, -20]
... ])
>>> y = np.array([0, 0, 1, 1, 2, 2])
>>> clf = OneVsRestClassifier(SVC()).fit(X, y)
>>> clf.predict([[-19, -20], [9, 9], [-5, 5]])
array([2, 0, 1])
Adapted by: Lee Prevost, https://github.com/leeprevost
"""
def __init__(self, estimator, *, n_jobs=None, reverse_classes=False):
self.estimator: BaseEstimator = estimator
self.n_jobs: int = n_jobs
self.reverse_classes: bool = reverse_classes
self.class_order: Iterable = []
self._class_ = None # private for class_
# validate estimator
if not self._has_predict_proba:
raise ValueError(
"Estimator {} does not have predict_proba method which is required for this classifier.".format(
self.estimator.__repr__()))
def fit(self, X, y):
"""Fit underlying estimators.
Parameters
----------
X : (sparse) array-like of shape (n_samples, n_features)
Data.
y : (sparse) array-like of shape (n_samples,) or (n_samples, n_classes)
Multi-class targets. An indicator matrix turns on multilabel
classification.
Returns
-------
self : object
Instance with fitted estimators_ as follows:
If X has n_classes_, (eg. classes: c0, c1, c2, c3 = 4)
Produce n-1 estimators each with the binary problem of classifying derived datasets as follows:
e1 - target > class0 (meaning > order) --> target = c1, c2, c3 (~ target != c0) (first class v rest - OvR)
e2 - target > class1 --> target = c2, c3 (~target != c0, c2) (second class v third, fourth (OvR)
e3 - target > class2 --> target = c3 (~ target != c0, c1, c2) (third class vs. fourth(OvO))
">" means a higher order than the target class.
"""
# @todo: keep? same as ovr?
# following improvised from
# https://towardsdatascience.com/simple-trick-to-train-an-ordinal-regression-with-any-classifier-6911183d2a3c
# by Muhammad
if self.class_order: # case to override everything
self.classes_ = self.class_order
# if order is given in the init, ignore reversed and ignore cat info
# validate class order and stop if invalid
# need validation of class order.
# raise error if not superset of class.
# warning if any missing classes not see during fit.
# if y has categorical info, capture it
elif hasattr(y, "cat"):
if y.cat.ordered: #has categories and its ordered
classes = y.cat.categories.to_numpy()
self.classes_ = classes # setter converts to np.array from index
else: #this is most likely path but handle other two cases above.
self.classes_ = np.sort(np.unique(y))
# ok, now order is set. Now, reverse it unless it was supplied
if self.reverse_classes and not self.class_order:
self.classes_ = self.classes_[::-1]
self.y_type_ = type_of_target(y)
if self.y_type_ is not "multiclass":
raise ValueError("This classifier expects target y to be multiclass. Got type: {}".format(self.y_type_))
# In cases where individual estimators are very fast to train setting
# n_jobs > 1 in can results in slower performance due to the overhead
# of spawning threads. See joblib issue #112.
if self.classes_.shape[0] > 2:
# for each k - 1 ordinal value we fit a binary classification problem
# @todo: question - should I allow for this to be reversed with kwargs in order to
# emphasize the positive class (eg. "hot" in cold < warm < hot three class problem)
# @todo: derived estimators: classes become imbalanced? how to balance classes?
# probable answer: make use of "class_weight" kwarg when fitting derived estimators?
y_derived, names = self._derived_ys(
y) # added helper to create vector of derived y data (of shape n_samples, n_classes-1)
self.estimators_ = Parallel(n_jobs=self.n_jobs)(
delayed(_fit_ovr_binary)(
self.estimator,
X,
y_d,
classes=[
"not %s" % self.classes_[i],
" or ".join(str(cls) for cls in self.classes_[i + 1:]),
],
)
for i, y_d in enumerate(y_derived.T)
) # create a binary estimator for each derived y
if hasattr(self.estimators_[0], "n_features_in_"):
self.n_features_in_ = self.estimators_[0].n_features_in_
if hasattr(self.estimators_[0], "feature_names_in_"):
self.feature_names_in_ = self.estimators_[0].feature_names_in_
return self
@available_if(_estimators_has("partial_fit"))
def partial_fit(self, X, y, classes=None):
"""Partially fit underlying estimators.
Should be used when memory is inefficient to train all data.
Chunks of data can be passed in several iteration.
Parameters
----------
X : (sparse) array-like of shape (n_samples, n_features)
Data.
y : (sparse) array-like of shape (n_samples,) or (n_samples, n_classes)
Multi-class targets. An indicator matrix turns on multilabel
classification.
classes : array, shape (n_classes, )
Classes across all calls to partial_fit.
Can be obtained via `np.unique(y_all)`, where y_all is the
target vector of the entire dataset.
This argument is only required in the first call of partial_fit
and can be omitted in the subsequent calls.
Returns
-------
self : object
Instance of partially fitted estimator.
"""
pass # for now bypass this and edit it later. @todo: implement partial_fit
'''
if _check_partial_fit_first_call(self, classes):
if not hasattr(self.estimator, "partial_fit"):
raise ValueError(
("Base estimator {0}, doesn't have partial_fit method").format(
self.estimator
)
)
self.estimators_ = [clone(self.estimator) for _ in range(self.n_classes_)]
# A sparse LabelBinarizer, with sparse_output=True, has been
# shown to outperform or match a dense label binarizer in all
# cases and has also resulted in less or equal memory consumption
# in the fit_ovr function overall.
self.label_binarizer_ = LabelBinarizer(sparse_output=True)
self.label_binarizer_.fit(self.classes_)
if len(np.setdiff1d(y, self.classes_)):
raise ValueError(
(
"Mini-batch contains {0} while classes " + "must be subset of {1}"
).format(np.unique(y), self.classes_)
)
# this is where we need n-1 targets from binarizer.
# y > Vi
Y = self.label_binarizer_.transform(y)
Y = Y.tocsc()
columns = (col.toarray().ravel() for col in Y.T)
self.estimators_ = Parallel(n_jobs=self.n_jobs)(
delayed(_partial_fit_binary)(estimator, X, column)
for estimator, column in zip(self.estimators_, columns)
)
if hasattr(self.estimators_[0], "n_features_in_"):
self.n_features_in_ = self.estimators_[0].n_features_in_
return self '''
def predict(self, X):
"""Predict multi-class targets using underlying estimators.
**estimator must have predict_proba method.
Parameters
----------
X : (sparse) array-like of shape (n_samples, n_features)
Data.
Returns
-------
y : (sparse) array-like of shape (n_samples,) or (n_samples, n_classes)
Predicted multi-class targets.
"""
check_is_fitted(self)
n_samples = _num_samples(X)
if self.y_type_ == "multiclass":
return self.classes_[np.argmax(self.predict_proba(X), axis=1)]
# need to rewrite the following if not "multiclass" or no predict_proba or want to use threshold
else:
# replaced elaborate else logic from OvR with NotImplementedError
raise NotImplementedError("This type of y target not implemented: type: ".format(self.y_type_))
@available_if(_estimators_has("predict_proba"))
def predict_proba(self, X):
"""Probability estimates.
The returned estimates for all classes are ordered by label of classes.
Note that in the multilabel case, each sample can have any number of
labels. This returns the marginal probability that the given sample has
the label in question. For example, it is entirely consistent that two
labels both have a 90% probability of applying to a given sample.
In the single label multiclass case, the rows of the returned matrix
sum to 1.
Parameters
----------
X : array-like of shape (n_samples, n_features)
Input data.
Returns
-------
T : (sparse) array-like of shape (n_samples, n_classes)
Returns the probability of the sample for each class in the model,
where classes are ordered as they are in `self.classes_`.
@todo: multilabel is untested
"""
check_is_fitted(self)
# Y[i, j] gives the probability that sample i has the label j.
# In the multi-label case, these are not disjoint.
Y = np.array([e.predict_proba(X)[:, 1] for e in self.estimators_]).T
if len(self.estimators_) == 1: # binary problem
# Only one estimator, but we still want to return probabilities
# for two classes.
Y = np.concatenate(((1 - Y), Y), axis=1)
predicted = Y
else:
predicted = self._ordinal_binary_to_class_array(Y)
if not self.multilabel_:
# Then, probabilities should be normalized to 1.
predicted /= np.sum(predicted, axis=1)[:, np.newaxis]
return predicted
@available_if(_estimators_has('decision_function'))
def decision_function(self, X):
"""Decision function for the OneVsRestClassifier.
Return the distance of each sample from the decision boundary for each
class. This can only be used with estimators which implement the
`decision_function` method.
Parameters
----------
X : array-like of shape (n_samples, n_features)
Input data.
Returns
-------
T : array-like of shape (n_samples, n_classes) or (n_samples,) for
binary classification.
Result of calling `decision_function` on the final estimator.
.. versionchanged:: 0.19
output shape changed to ``(n_samples,)`` to conform to
scikit-learn conventions for binary classification.
"""
check_is_fitted(self)
if len(self.estimators_) == 1:
return self.estimators_[0].decision_function(X)
else:
Y = np.array([e.predict_proba(X) for e in self.estimators_]).T
decision = self._ordinal_binary_to_class_array(Y)
return decision
def _ordinal_binary_to_class_array(self, Y):
predicted = []
pr_name = "Pr(y={})"
for i, cls in enumerate(self.classes_):
pr_names = "Pr"
if i == 0: # first pass
# Pr(V1) = 1 − Pr(Target > V1)
predicted.append((pr_name.format(cls), 1 - Y[:, 0])) # first class
elif cls == self.classes_[-1]: # last pass
# Pr(Vk) = Pr(Target > Vk−1)
predicted.append((pr_name.format(cls), Y[:, -1])) # last class
elif i > 0: # middle passes, need qualifier so it doesn't overwrite last class. this shouldn't exec on i=0 and last pass.
# Pr(Vi) = Pr(Target > Vi−1) − Pr(Target > Vi) , 1 < i < k
predicted.append((pr_name.format(cls), Y[:, i - 1] - Y[:, i])) # middle classes
self.ordinal_prob_names_ = [name for name, _ in predicted]
predicted = np.vstack(list(prob for _, prob in predicted)).T
return predicted
def _derived_ys(self, y):
"""private function that generates n_classes - 1 derived y datasets which iterate through
classes_ with a ptr and does comparison to remaining classes pointed to beyond current class
eg. np.isin(y, classes_[ptr:])
returns array of probabilities, names or arrays
consider classes_ = 0, 1, 2, 3, 4
4 estimators (n_classes -1)
ovr(emaining)
binary estimators derived ys (0|1)
e1 (y>c0) y(0|1,2,3,4)
e2 (y>c1) y(1|2,3,4)
e3 (y>c2) y(2|3,4)
e4 (y>c3) y(3|4)
I found the Ordinal Classifier white paper to be very difficult to follow until I understood:
Prob(target > cool) ~ y cool|warm,hot
"""
derived = []
names = []
for i in range(len(self.classes_) - 1):
ptr = i + 1 # pts to start ndx of remaining classes
# one class vs. remaining classes
class_name = self.classes_[i] # 'one' class name
remaining_classes = self.classes_[ptr:] # r - remaining classes
y_ = np.isin(y, remaining_classes) * 1
derived.append(y_)
names.append("V{}: y>class({})".format(ptr, class_name))
return np.vstack(derived).T, np.array(names)
@property
def multilabel_(self):
"""Whether this is a multilabel classifier."""
return self.y_type_.startswith("multilabel")
@property
def n_classes_(self):
"""Number of classes."""
return len(self.classes_)
# TODO: Remove coef_ attribute in 1.1
# mypy error: Decorated property not supported
@deprecated( # type: ignore
"Attribute `coef_` was deprecated in "
"version 0.24 and will be removed in 1.1 (renaming of 0.26). "
"If you observe this warning while using RFE "
"or SelectFromModel, use the importance_getter "
"parameter instead."
)
@property
def coef_(self):
check_is_fitted(self)
if not hasattr(self.estimators_[0], "coef_"):
raise AttributeError("Base estimator doesn't have a coef_ attribute.")
coefs = [e.coef_ for e in self.estimators_]
if sp.issparse(coefs[0]):
return sp.vstack(coefs)
return np.vstack(coefs)
# TODO: Remove intercept_ attribute in 1.1
# mypy error: Decorated property not supported
@deprecated( # type: ignore
"Attribute `intercept_` was deprecated in "
"version 0.24 and will be removed in 1.1 (renaming of 0.26). "
"If you observe this warning while using RFE "
"or SelectFromModel, use the importance_getter "
"parameter instead."
)
@property
def intercept_(self):
check_is_fitted(self)
if not hasattr(self.estimators_[0], "intercept_"):
raise AttributeError("Base estimator doesn't have an intercept_ attribute.")
return np.array([e.intercept_.ravel() for e in self.estimators_])
# TODO: Remove in 1.1
# mypy error: Decorated property not supported
@deprecated( # type: ignore
"Attribute `_pairwise` was deprecated in "
"version 0.24 and will be removed in 1.1 (renaming of 0.26)."
)
@property
def _pairwise(self):
"""Indicate if wrapped estimator is using a precomputed Gram matrix"""
return getattr(self.estimator, "_pairwise", False)
def _more_tags(self):
"""Indicate if wrapped estimator is using a precomputed Gram matrix"""
return {"pairwise": _safe_tags(self.estimator, key="pairwise")}
@property
def _has_decision_function(self):
return hasattr(self.estimator, "decision_function")
@property
def _has_predict_proba(self):
return hasattr(self.estimator, "predict_proba")
@property
def class_(self):
return self._class_
@class_.setter
def class_(self, iterable):
self._class_ = np.array(iterable)