Principal component analysis is a mathematical transformation that can be understood in two parts:
- the transformation maps multivariable data (Nold dimensions) into a new coordinate system (Nnew dimensions) with minimal loss of information.
- data projected on the first dimension of the new coordinate system, also known as the first principal component, has the greatest variance. Data projected on the second dimension of the new coordinate system has the second greatest variance.
PCA is useful as a feature extraction method because it can reduce complex multivariable data to fewer dimensions (e.g. 100 dimensions to 10 dimensions) without loss of important characteristic information.
- 1. Alpaydın E. Introduction to Machine Learning. The Massachusetts Institute of Technology Press. Cambridge, Massachusetts, USA.
Related Radiopaedia articles
- artificial intelligence (AI)
- imaging data sets
- computer-aided diagnosis (CAD)
machine learning (overview)
- machine learning processes
- machine learning models
- visualizing and understanding neural networks
- common data preparation/preprocessing steps
- DICOM to bitmap conversion
- principal component analysis
- training, testing and validation datasets
- loss function
- optimization algorithms
- stochastic gradient descent
- momentum (Nesterov)
- linear and quadratic
- batch normalization
- natural language processing
- rule-based expert systems