Pipeline ANOVA SVM¶
This example shows how a feature selection can be easily integrated within a machine learning pipeline.
We also show that you can easily introspect part of the pipeline.
We will start by generating a binary classification dataset. Subsequently, we will divide the dataset into two subsets.
from sklearn.datasets import make_classification from sklearn.model_selection import train_test_split X, y = make_classification( n_features=20, n_informative=3, n_redundant=0, n_classes=2, n_clusters_per_class=2, random_state=42, ) X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=42)
A common mistake done with feature selection is to search a subset of
discriminative features on the full dataset instead of only using the
training set. The usage of scikit-learn
prevents to make such mistake.
Here, we will demonstrate how to build a pipeline where the first step will be the feature selection.
fit on the training data, a subset of feature will be selected
and the index of these selected features will be stored. The feature selector
will subsequently reduce the number of feature and pass this subset to the
classifier which will be trained.
Pipeline(steps=[('selectkbest', SelectKBest(k=3)), ('linearsvc', LinearSVC())])In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.
Pipeline(steps=[('selectkbest', SelectKBest(k=3)), ('linearsvc', LinearSVC())])
Once the training accomplished, we can predict on new unseen samples. In this case, the feature selector will only select the most discriminative features based on the information stored during training. Then, the data will be passed to the classifier which will make the prediction.
Here, we report the final metrics via a classification report.
precision recall f1-score support 0 0.92 0.80 0.86 15 1 0.75 0.90 0.82 10 accuracy 0.84 25 macro avg 0.84 0.85 0.84 25 weighted avg 0.85 0.84 0.84 25
Be aware that you can inspect a step in the pipeline. For instance, we might be interested about the parameters of the classifier. Since we selected three features, we expect to have three coefficients.
array([[0.75790919, 0.27158706, 0.26109741]])
However, we do not know which features where selected from the original dataset. We could proceed by several manner. Here, we will inverse the transformation of these coefficients to get information about the original space.
array([[0. , 0. , 0.75790919, 0. , 0. , 0. , 0. , 0. , 0. , 0.27158706, 0. , 0. , 0. , 0. , 0. , 0. , 0. , 0. , 0. , 0.26109741]])
We can see that the first three features where the selected features by the first step.
Total running time of the script: ( 0 minutes 0.008 seconds)