Robust Principal Components analysisRobust Principal Components Analysis
RPCA is splitting the data in two matrices:
- Low Rank Matrix: Describes the main features in the data, noise and outliers removed
- Sparse Matrix: Consisting of noise AND signals! Makes it quite easy to detect signals/special causes. Can be used e.g. to investigate life time of engines or servo drives.
Method has in general a higher resolution to find groups than standard PCA, apart from this the Low Rank Matrix is more efficient in Feature Extraction than PCA.
Statistics and Analytics
