ordiplot

ordiplot(result, options?): any

Defined in: plot/ordiplot.js:79

Generate unified ordination plot configuration Works with PCA, LDA, and RDA results

Parameters

result

any

Ordination result (from PCA, LDA, or RDA)

options?

Configuration options

axis1

number = 1

First axis to plot (default: 1)

axis2

number = 2

Second axis to plot (default: 2)

color

any = null

Point color scale for grouped points, merged into the Plot color scale: { range, scheme, domain, legend, label }. E.g. { range: ["#111","#888"] } for greyscale, or { scheme: "Observable10" }.

colorBy

any = null

Group values for points: an array, any iterable (e.g. an Arquero column), a { data, column } descriptor, or the name of a column in the data the model was fit on (requires a declarative fit({ data, … }))

height

number = 400

Plot height (default: 400)

labelNudge

number = 1

Multiplier on the radial distance a loading/ predictor label sits past its arrow tip (default 1). Larger pushes labels further out from the tips before de-collision.

labelRepel

boolean = true

Run ggrepel-style overlap removal on the arrow labels (default true). Set false to place each label at its tip with no de-collision.

labels

any = null

Labels for points (same accepted forms as colorBy)

loadingColor

string = null

Stroke for loading arrows (default: ‘red’, or ‘blue’ for an RDA response triplot).

loadingFactor

number = 0

Multiplier applied to loading vectors. Default 0 = AUTO: fit the longest vector to 90% of the score cloud’s radius, so arrows and points are readable against each other. Set a number to override — 1 draws the loadings at their raw length, which for PCA is usually far larger than the scores (site scores are normalized to unit column norm, hence of order 1/sqrt(n), while loadings stay of order 1) and leaves the points crushed into a dot at the origin.

loadingScale

number = 3

Constant applied to loading vectors before loadingFactor (default: 3). Under the default auto-scaling it has no effect: the auto-fit normalizes by the longest vector, which cancels any constant prefactor. It only bites when loadingFactor is set explicitly.

loadingTextColor

string = null

Fill for loading labels (default: ‘darkred’, or ‘darkblue’ for RDA).

minLoadingContribution

number = 0

Hide loading vectors whose contribution to the two displayed axes is below this fraction (0-1) of the total, i.e. squared vector length / summed squared length. Default 0 shows every vector at its true relative length. Use e.g. 0.02 to drop near-zero vectors that only clutter the plot, keeping magnitudes honest (vectors are NOT rescaled - negligible ones are removed, not inflated). On an RDA this governs the response arrows, and minPredictorContribution the predictor ones.

minPredictorContribution

number = null

Same filter for RDA predictor arrows (default: null, inherits minLoadingContribution). Each group is measured against its own summed length, and an RDA typically has a handful of responses against many predictors, so one threshold rarely suits both: a response competing with three others clears 2% easily, while the same arrow among twenty predictors may not. Set this to declutter the predictors while keeping every response visible - a response dropping out is usually a finding (these axes carry none of it), not clutter.

pointColor

string = 'steelblue'

Fill for points when there is no colorBy (default: ‘steelblue’).

pointRadius

number = 4

Point radius (default: 4).

predictorColor

string = null

Stroke for RDA predictor arrows (default: ‘red’). Set e.g. ‘#111’ for a greyscale triplot.

predictorFactor

number = null

Multiplier for predictor arrows (RDA only, default: inherits loadingFactor, so auto as well; set a number to override)

predictorTextColor

string = null

Fill for RDA predictor labels (default: inherits predictorColor, else ‘darkred’).

showCentroids

boolean = false

Show class centroids (LDA only)

showConvexHulls

boolean = false

Show convex hulls around groups (optional)

showLoadings

boolean = true

Show loading vectors (PCA/RDA only)

symbolBy

boolean = false

Also encode groups by symbol (shape), not colour alone - recommended for greyscale / colour-vision-deficiency safety.

type

string = null

Type of ordination (‘pca’, ‘lda’, ‘rda’) - auto-detected if not specified

width

number = 640

Plot width (default: 640)

Returns

any

Plot configuration