PrinCompOptions

Sets options for principal component analysis. The default options are also the values in PrinCompOptions.init.

struct PrinCompOptions {

}

Members

absError double absError;: Absolute error at which to stop the optimization procedure. Default: 1e-5
destructive bool destructive;: Overwrite input matrix instead of copying. Ignored if the matrix passed in does not have assignable, lvalue elements and centering or scaling is enabled. Default value: false.
maxIter uint maxIter;: Maximum iterations for the optimization procedure. After this many iterations, the algorithm gives up and calls teh solution "good enough" no matter what. For exploratory analyses, "good enough" solutions can be had fast sometimes by making this value small. Default: uint.max
relError double relError;: Relative error at which to stop the optimization procedure. Default: 1e-4
transpose bool transpose;: Effectively transpose the matrix. If enabled, treat each column as a data points and each row as a dimension. If disabled, do the opposite. Note that, if this is enabled, each row will be scaled and centered, not each column. Default value: false.
unitVariance bool unitVariance;: Scale each column to unit variance. Note that, if this option is set to true, zeroMean is ignored and the mean of each column is set to zero even if zeroMean is false. Default value: false.
zeroMean bool zeroMean;: Center each column to zero mean. Default value: true.