Note
Go to the end to download the full example code.
Fitting model on imbalanced datasets and how to fight bias#
This example illustrates the problem induced by learning on datasets having imbalanced classes. Subsequently, we compare different approaches alleviating these negative effects.
# Authors: Guillaume Lemaitre <g.lemaitre58@gmail.com>
# License: MIT
print(__doc__)
Problem definition#
We are dropping the following features:
“fnlwgt”: this feature was created while studying the “adult” dataset. Thus, we will not use this feature which is not acquired during the survey.
“education-num”: it is encoding the same information than “education”. Thus, we are removing one of these 2 features.
The “adult” dataset as a class ratio of about 3:1
class
<=50K 37155
>50K 11687
Name: count, dtype: int64
This dataset is only slightly imbalanced. To better highlight the effect of learning from an imbalanced dataset, we will increase its ratio to 30:1
from imblearn.datasets import make_imbalance
ratio = 30
df_res, y_res = make_imbalance(
df,
y,
sampling_strategy={classes_count.idxmin(): classes_count.max() // ratio},
)
y_res.value_counts()
class
<=50K 37155
>50K 1238
Name: count, dtype: int64
We will perform a cross-validation evaluation to get an estimate of the test score.
As a baseline, we could use a classifier which will always predict the majority class independently of the features provided.
from sklearn.dummy import DummyClassifier
from sklearn.model_selection import cross_validate
dummy_clf = DummyClassifier(strategy="most_frequent")
scoring = ["accuracy", "balanced_accuracy"]
cv_result = cross_validate(dummy_clf, df_res, y_res, scoring=scoring)
print(f"Accuracy score of a dummy classifier: {cv_result['test_accuracy'].mean():.3f}")
Accuracy score of a dummy classifier: 0.968
Instead of using the accuracy, we can use the balanced accuracy which will take into account the balancing issue.
print(
"Balanced accuracy score of a dummy classifier: "
f"{cv_result['test_balanced_accuracy'].mean():.3f}"
)
Balanced accuracy score of a dummy classifier: 0.500
Strategies to learn from an imbalanced dataset#
We will use a dictionary and a list to continuously store the results of our experiments and show them as a pandas dataframe.
Dummy baseline#
Before to train a real machine learning model, we can store the results
obtained with our DummyClassifier.
import pandas as pd
index += ["Dummy classifier"]
cv_result = cross_validate(dummy_clf, df_res, y_res, scoring=scoring)
scores["Accuracy"].append(cv_result["test_accuracy"].mean())
scores["Balanced accuracy"].append(cv_result["test_balanced_accuracy"].mean())
df_scores = pd.DataFrame(scores, index=index)
df_scores
Linear classifier baseline#
We will create a machine learning pipeline using a
LogisticRegression classifier. In this regard,
we will need to one-hot encode the categorical columns and standardized the
numerical columns before to inject the data into the
LogisticRegression classifier.
First, we define our numerical and categorical pipelines.
from sklearn.impute import SimpleImputer
from sklearn.pipeline import make_pipeline
from sklearn.preprocessing import OneHotEncoder, StandardScaler
num_pipe = make_pipeline(
StandardScaler(), SimpleImputer(strategy="mean", add_indicator=True)
)
cat_pipe = make_pipeline(
SimpleImputer(strategy="constant", fill_value="missing"),
OneHotEncoder(handle_unknown="ignore"),
)
Then, we can create a preprocessor which will dispatch the categorical columns to the categorical pipeline and the numerical columns to the numerical pipeline
from sklearn.compose import make_column_selector as selector
from sklearn.compose import make_column_transformer
preprocessor_linear = make_column_transformer(
(num_pipe, selector(dtype_include="number")),
(cat_pipe, selector(dtype_include="category")),
n_jobs=2,
)
Finally, we connect our preprocessor with our
LogisticRegression. We can then evaluate our
model.
from sklearn.linear_model import LogisticRegression
lr_clf = make_pipeline(preprocessor_linear, LogisticRegression(max_iter=1000))
We can see that our linear model is learning slightly better than our dummy baseline. However, it is impacted by the class imbalance.
We can verify that something similar is happening with a tree-based model
such as RandomForestClassifier. With this type of
classifier, we will not need to scale the numerical data, and we will only
need to ordinal encode the categorical data.
from sklearn.ensemble import RandomForestClassifier
from sklearn.preprocessing import OrdinalEncoder
num_pipe = SimpleImputer(strategy="mean", add_indicator=True)
cat_pipe = make_pipeline(
SimpleImputer(strategy="constant", fill_value="missing"),
OrdinalEncoder(handle_unknown="use_encoded_value", unknown_value=-1),
)
preprocessor_tree = make_column_transformer(
(num_pipe, selector(dtype_include="number")),
(cat_pipe, selector(dtype_include="category")),
n_jobs=2,
)
rf_clf = make_pipeline(
preprocessor_tree, RandomForestClassifier(random_state=42, n_jobs=2)
)
The RandomForestClassifier is as well affected by
the class imbalanced, slightly less than the linear model. Now, we will
present different approach to improve the performance of these 2 models.
Use class_weight#
Most of the models in scikit-learn have a parameter class_weight. This
parameter will affect the computation of the loss in linear model or the
criterion in the tree-based model to penalize differently a false
classification from the minority and majority class. We can set
class_weight="balanced" such that the weight applied is inversely
proportional to the class frequency. We test this parametrization in both
linear model and tree-based model.
lr_clf.set_params(logisticregression__class_weight="balanced")
index += ["Logistic regression with balanced class weights"]
cv_result = cross_validate(lr_clf, df_res, y_res, scoring=scoring)
scores["Accuracy"].append(cv_result["test_accuracy"].mean())
scores["Balanced accuracy"].append(cv_result["test_balanced_accuracy"].mean())
df_scores = pd.DataFrame(scores, index=index)
df_scores
rf_clf.set_params(randomforestclassifier__class_weight="balanced")
index += ["Random forest with balanced class weights"]
cv_result = cross_validate(rf_clf, df_res, y_res, scoring=scoring)
scores["Accuracy"].append(cv_result["test_accuracy"].mean())
scores["Balanced accuracy"].append(cv_result["test_balanced_accuracy"].mean())
df_scores = pd.DataFrame(scores, index=index)
df_scores
We can see that using class_weight was really effective for the linear
model, alleviating the issue of learning from imbalanced classes. However,
the RandomForestClassifier is still biased toward
the majority class, mainly due to the criterion which is not suited enough to
fight the class imbalance.
Resample the training set during learning#
Another way is to resample the training set by under-sampling or
over-sampling some of the samples. imbalanced-learn provides some samplers
to do such processing.
from imblearn.pipeline import make_pipeline as make_pipeline_with_sampler
from imblearn.under_sampling import RandomUnderSampler
lr_clf = make_pipeline_with_sampler(
preprocessor_linear,
RandomUnderSampler(random_state=42),
LogisticRegression(max_iter=1000),
)
index += ["Under-sampling + Logistic regression"]
cv_result = cross_validate(lr_clf, df_res, y_res, scoring=scoring)
scores["Accuracy"].append(cv_result["test_accuracy"].mean())
scores["Balanced accuracy"].append(cv_result["test_balanced_accuracy"].mean())
df_scores = pd.DataFrame(scores, index=index)
df_scores
rf_clf = make_pipeline_with_sampler(
preprocessor_tree,
RandomUnderSampler(random_state=42),
RandomForestClassifier(random_state=42, n_jobs=2),
)
index += ["Under-sampling + Random forest"]
cv_result = cross_validate(rf_clf, df_res, y_res, scoring=scoring)
scores["Accuracy"].append(cv_result["test_accuracy"].mean())
scores["Balanced accuracy"].append(cv_result["test_balanced_accuracy"].mean())
df_scores = pd.DataFrame(scores, index=index)
df_scores
Applying a random under-sampler before the training of the linear model or random forest, allows to not focus on the majority class at the cost of making more mistake for samples in the majority class (i.e. decreased accuracy).
We could apply any type of samplers and find which sampler is working best on the current dataset.
Instead, we will present another way by using classifiers which will apply sampling internally.
Use of specific balanced algorithms from imbalanced-learn#
We already showed that random under-sampling can be effective on decision
tree. However, instead of under-sampling once the dataset, one could
under-sample the original dataset before to take a bootstrap sample. This is
the base of the imblearn.ensemble.BalancedRandomForestClassifier and
BalancedBaggingClassifier.
from imblearn.ensemble import BalancedRandomForestClassifier
rf_clf = make_pipeline(
preprocessor_tree,
BalancedRandomForestClassifier(
sampling_strategy="all",
replacement=True,
bootstrap=False,
random_state=42,
n_jobs=2,
),
)
The performance with the
BalancedRandomForestClassifier is better than
applying a single random under-sampling. We will use a gradient-boosting
classifier within a BalancedBaggingClassifier.
from sklearn.ensemble import HistGradientBoostingClassifier
from imblearn.ensemble import BalancedBaggingClassifier
bag_clf = make_pipeline(
preprocessor_tree,
BalancedBaggingClassifier(
estimator=HistGradientBoostingClassifier(random_state=42),
n_estimators=10,
random_state=42,
n_jobs=2,
),
)
index += ["Balanced bag of histogram gradient boosting"]
cv_result = cross_validate(bag_clf, df_res, y_res, scoring=scoring)
scores["Accuracy"].append(cv_result["test_accuracy"].mean())
scores["Balanced accuracy"].append(cv_result["test_balanced_accuracy"].mean())
df_scores = pd.DataFrame(scores, index=index)
df_scores
This last approach is the most effective. The different under-sampling allows to bring some diversity for the different GBDT to learn and not focus on a portion of the majority class.
Total running time of the script: (0 minutes 37.257 seconds)
Estimated memory usage: 295 MB