The `MatSignalExtractJP` function performs signal extraction from a data matrix to estimate the signal rank, signal row space, corresponding perturbation angles, and noise matrix. It adjusts the signals based on random direction angles and uses bootstrap estimation to refine the results.
Usage
MatSignalExtractJP(
X,
matName = NULL,
nsim = 400,
colCent = F,
rowCent = F,
cull = 0.382,
percentile = 0.5,
noiselvl = NULL
)Arguments
- X
A numeric matrix (d x n) containing the data to be analyzed. Must not contain any `NA` values.
- matName
A string representing the name of the matrix, used for logging purposes.
- nsim
An integer specifying the number of bootstrap samples to be used during the estimation process.
- colCent
A logical value indicating whether to center the columns of the matrix. Default is `FALSE`.
- rowCent
A logical value indicating whether to center the rows of the matrix. Default is `FALSE`.
- cull
A numeric value controlling the culling process during signal extraction. Default is `0.5`.
- percentile
A numeric value representing the percentile for estimating the noise level.
- noiselvl
Optional. A noise level parameter that can be either a numeric value or 'ks' to use the `ksOpt` function for noise level estimation. If not provided, an optimal shrinkage method will be used.
Value
A list containing:
- VBar
Matrix representing the adjusted signal row space.
- UBar
Matrix representing the adjusted signal column space.
- phiBar
Numeric value of the adjusted perturbation angle.
- psiBar
Numeric value of the loadings perturbation angle.
- rBar
Integer representing the adjusted signal rank.
- EHat
Matrix representing the estimated noise.
- singVals
Vector of singular values before shrinkage.
- singValsHat
Vector of singular values after shrinkage.
- rSteps
List of steps used in signal rank adjustment.
- VVHatCacheBar
List of matrices from bootstrap validation steps.
- UUHatCacheBar
List of matrices from bootstrap validation steps.