6.5. Unsupervised dimensionality reduction¶
If your number of features is high, it may be useful to reduce it with an
unsupervised step prior to supervised steps. Many of the
Unsupervised learning methods implement a transform
method that
can be used to reduce the dimensionality. Below we discuss two specific
example of this pattern that are heavily used.
Pipelining
The unsupervised data reduction and the supervised estimator can be chained in one step. See Pipeline: chaining estimators.
6.5.1. PCA: principal component analysis¶
decomposition.PCA
looks for a combination of features that
capture well the variance of the original features. See Decomposing signals in components (matrix factorization problems).
6.5.2. Random projections¶
The module: random_projection
provides several tools for data
reduction by random projections. See the relevant section of the
documentation: Random Projection.
6.5.3. Feature agglomeration¶
cluster.FeatureAgglomeration
applies
Hierarchical clustering to group together features that behave
similarly.
Feature scaling
Note that if features have very different scaling or statistical
properties, cluster.FeatureAgglomeration
may not be able to
capture the links between related features. Using a
preprocessing.StandardScaler
can be useful in these settings.