Chapter 8. Dimensionality Reduction
Building a useful predictive model requires analyzing an appropriate number of observations (or cases). This number will vary, based upon your project or your objective. Strictly speaking, the more variations (not necessarily the more data) analyzed, the better the outcome or results of the model.
This chapter will discuss the concept of reducing the size or amount of the data being observed without affecting the outcome of the analysis (or the success of the project), through various common approaches such as correlation analysis, principal component analysis, independent component analysis, common factor analysis, and non-negative matrix factorization.
Let us begin by clarifying what is meant by dimensional ...