Online PCA
Block Catalogue / Analytics
Streaming Principal Component Analysis — incrementally tracks the top k principal directions (Oja/Sanger update with periodic QR re-orthonormalisation) and projects each sample onto them.
Input and output
| Property | Value |
|---|---|
| Type | onlinepca, mosaic.models.onlinepca |
| Inputs | exactly 1 |
| Consumes | Vector<double> or Matrix<double> (row-wise) |
| Publishes | Vector<double> of length k, one per observation |
Parameters
| # | Name | Type | Default | Meaning |
|---|---|---|---|---|
| 0 | k | int | 2 | Components; must be 2 or 3 |
| 1 | eta0 | double | 0.2 | Learning rate; decays as eta0/√n |
| 2 | reorthEvery | int | 100 | Re-orthonormalisation interval |
| 3 | minStableCount | int | 500 | Samples before IsStable flips |
Requirements and use
Choose a component count that fits the feature dimension and allow adaptation to settle. Component directions can change as the online model learns.
For parameter positions and connection rules, see Pipeline Configuration. For capturing output, see Recording.
Implementation
API reference. Model defaults above describe configuration loading; palette hints may expose a subset or select different defaults.