R: Identify the Optimal Contrastive and Penalty Parameters

fitGrid {scPCA}

R Documentation

Identify the Optimal Contrastive and Penalty Parameters

Description

This function is used to automatically select the optimal contrastive parameter and L1 penalty term for scPCA based on a clustering algorithm and average silhouette width.

Usage

fitGrid(
  target,
  target_valid = NULL,
  center,
  scale,
  c_contrasts,
  contrasts,
  alg,
  penalties,
  n_eigen,
  clust_method = c("kmeans", "pam", "hclust"),
  n_centers,
  max_iter = 10,
  linkage_method = "complete",
  clusters = NULL,
  eigdecomp_tol = 1e-10,
  eigdecomp_iter = 1000
)

Arguments

`target`	The target (experimental) data set, in a standard format such as a `data.frame` or `matrix`.
`target_valid`	A holdout set of the target (experimental) data set, in a standard format such as a `data.frame` or `matrix`. `NULL` by default but used by `cvSelectParams` for cross-validated selection of the contrastive and penalization parameters.
`center`	A `logical` indicating whether the target and background data sets should be centered to mean zero.
`scale`	A `logical` indicating whether the target and background data sets should be scaled to unit variance.
`c_contrasts`	A `list` of contrastive covariances.
`contrasts`	A `numeric` vector of the contrastive parameters used to compute the contrastive covariances.
`alg`	A `character` indicating the SPCA algorithm used to sparsify the contrastive loadings. Currently supports `iterative` for the Zou et al. (2006) implemententation, `var_proj` for the non-randomized Erichson et al. (2018) solution, and `rand_var_proj` for the randomized Erichson et al. (2018) result.
`penalties`	A `numeric` vector of the penalty terms.
`n_eigen`	A `numeric` indicating the number of eigenvectors to be computed.
`clust_method`	A `character` specifying the clustering method to use for choosing the optimal constrastive parameter. Currently, this is limited to either k-means, partitioning around medoids (PAM), and hierarchical clustering. The default is k-means clustering.
`n_centers`	A `numeric` giving the number of centers to use in the clustering algorithm.
`max_iter`	A `numeric` giving the maximum number of iterations to be used in k-means clustering, defaulting to 10.
`linkage_method`	A `character` specifying the agglomerative linkage method to be used if `clust_method = "hclust"`. The options are `ward.D2`, `single`, `complete`, `average`, `mcquitty`, `median`, and `centroid`. The default is `complete`.
`clusters`	A `numeric` vector of cluster labels for observations in the `target` data. Defaults to `NULL`, but is otherwise used to identify the optimal set of hyperparameters when fitting the scPCA and the automated version of cPCA.
`eigdecomp_tol`	A `numeric` providing the level of precision used by eigendecompositon calculations. Defaults to `1e-10`.
`eigdecomp_iter`	A `numeric` indicating the maximum number of interations performed by eigendecompositon calculations. Defaults to `1000`.

Value

A list similar to that output by prcomp:

rotation - the matrix of variable loadings
x - the rotated data, centred and scaled, if requested, data multiplied by the rotation matrix
contrast - the optimal contrastive parameter
penalty - the optimal L1 penalty term

References

Erichson NB, Zeng P, Manohar K, Brunton SL, Kutz JN, Aravkin AY (2018). “Sparse Principal Component Analysis via Variable Projection.” ArXiv, abs/1804.00341.

Zou H, Hastie T, Tibshirani R (2006). “Sparse principal component analysis.” Journal of computational and graphical statistics, 15(2), 265–286.

[Package scPCA version 1.6.2 Index]