dimensionality reduction
when dimensionality is very high occupation of the space become sparse and discrimination based on distance becomes ineffective
the scope of this processing is to:
- avoid this problem called the curse of dimensionality
- reduce noise on data
- reduce time and memory complexity of the mining algorithms
- improve visualization
pca
procedure to extract important variables from the dataset and reduce dimensionality. The new dataset will contains the attributes that capture most of the data variation
the PCA algorithm uses singular value decomposition to find the attributes that capture most of the data variation and reduce the number of dimensions in the dataset to make it more suitable for ML processes