MvaMethod R6 Class

Base class for multivariate analysis methods. Subclass this to implement new methods (e.g., CPCA, CCA, ICA). See PlsMethod for an example.

Public fields

name

Short identifier (e.g., "pls_task", "cpca")

label

Human-readable label (e.g., "Task PLS", "Constrained PCA")

family

Method family grouping variants (e.g., "pls", "cpca")

capabilities

Named list declaring what this method needs and produces

Methods

Public methods

• MvaMethod$new()

• MvaMethod$fit()

• MvaMethod$permutation_fit()

• MvaMethod$bootstrap_fit()

• MvaMethod$project_scores()

• MvaMethod$align_to_reference()

• MvaMethod$validate_spec()

• MvaMethod$configure()

• MvaMethod$ui_config()

• MvaMethod$clone()

---

Method new()

Create a new MvaMethod instance

Usage

MvaMethod$new(name, label, family, capabilities = list())

Arguments

name

Short identifier

label

Human-readable label

family

Method family

capabilities

Named list of capability flags

---

Method fit()

Fit the method to data. Must return an mva_decomposition.

Usage

MvaMethod$fit(spec, progress = TRUE)

Arguments

spec

An mva_spec or pls_spec object

progress

Logical, show progress

Returns

An mva_decomposition object

---

Method permutation_fit()

Fit on permuted data for significance testing. Default: permute rows of data matrices, then call fit().

Usage

MvaMethod$permutation_fit(spec, perm_order)

Arguments

spec

An mva_spec or pls_spec object

perm_order

Integer vector of permuted row indices

Returns

An mva_decomposition object

---

Method bootstrap_fit()

Fit on bootstrapped data for stability estimation. Default: resample rows, then call fit().

Usage

MvaMethod$bootstrap_fit(spec, boot_order)

Arguments

spec

An mva_spec or pls_spec object

boot_order

Integer vector of bootstrap row indices

Returns

An mva_decomposition object

---

Method project_scores()

Project held-out observations into the fitted score space. Subclasses should use only training-fit artifacts stored on result.

Usage

MvaMethod$project_scores(result, spec, type = "feature", progress = FALSE)

Arguments

result

A fitted mva_result

spec

Held-out mva_spec or pls_spec

type

Score space to project into

progress

Logical, show progress

Returns

Matrix of held-out scores, or a named list when type = "both"

---

Method align_to_reference()

Align bootstrap decomposition to reference (e.g., Procrustes). Default: sign-flip columns to maximize agreement.

Usage

MvaMethod$align_to_reference(boot_decomp, ref_decomp)

Arguments

boot_decomp

An mva_decomposition from bootstrap

ref_decomp

The reference mva_decomposition

Returns

Aligned mva_decomposition

---

Method validate_spec()

Validate that a spec has everything this method needs.

Usage

MvaMethod$validate_spec(spec)

Arguments

spec

An mva_spec or pls_spec object

Returns

TRUE invisibly, or stops with error

---

Method configure()

Apply method-specific configuration to a spec. Override to handle method-specific parameters.

Usage

MvaMethod$configure(spec, ...)

Arguments

spec

An mva_spec or pls_spec object

...

Method-specific parameters

Returns

Modified spec

---

Method ui_config()

UI metadata for the Shiny app. Override to customize.

Usage

MvaMethod$ui_config()

Returns

Named list with analysis_stages, explore_panels, inspector_metrics

---

Method clone()

The objects of this class are cloneable with this method.

Usage

MvaMethod$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.