ABSTRACT
This chapter reviews the glitches associated while handling this huge high-dimensional data. It discusses the techniques applied to reduce the dimension of data. The reduction in training features results in lesser assumptions made by the ML model and makes it simple. Dimensionality reduction techniques remove some of the unimportant features. Hence, ML model accuracy improves with the reduction in misleading data. ML techniques have proven applications to various fields starting from finance to mechanical engineering. Huge data dimension is a curse to the ML techniques, and dimensionality reduction techniques are possible solutions to this curse. Dimensionality reduction techniques such as forward/backward sequential removal techniques analyse the impact of one feature at a time. However, variance and covariance have an indirect impact on the performance of these techniques. Medical databases are high-dimensional databases. Classification analysis may generate inaccurate results on training dataset with irrelevant medical features.
