Definition - What does Dimensionality Reduction mean?
Dimensionality reduction is a series of techniques in machine learning and statistics to reduce the number of random variables to consider. It involves feature selection and feature extraction. Dimensionality reduction makes analyzing data much easier and faster for machine learning algorithms without extraneous variables to process, making machine learning algorithms faster and simpler in turn.
Techopedia explains Dimensionality Reduction
Dimensionality reduction attempts to reduce the number of random variables in data. A K-nearest-neighbors approach is often used. Dimensionality reduction techniques are divided into two major categories: feature selection and feature extraction.
Feature selection techniques find a smaller subset of a many-dimensional data set to create a data model. The major strategies for feature set are filter, wrapper (using a predictive model) and embedded, which perform feature selection while building a model.
Feature extraction involves transforming high-dimensional data into spaces of fewer dimensions. Methods include principal component analysis, kernel PCA, graph-based kernel PCA, linear discriminant analysis and generalized discriminant analysis.