A4.2.3 Dimensionality Reduction
In machine learning, dimensionality means the number of features in a dataset: one feature is one column. More features sound like more information, and up to a point they are. Beyond that point, adding features starts to hurt the model rather than help it, damaging accuracy, slowing training, and making the data harder to analyse. The problems that appear are known collectively as the curse of dimensionality, and...