ilmQA

Multivariate outlier detection based on PCA of QA statistics

Bundled with arrayMvout · exported, loads via data()

Computed over every cell in the object, reading it as one vector.

The mean of each row, and of each column, summarised. Reading every cell as one vector gives a matrix and its transpose the same answer; these do not.

The column means vary far more than the row means, so what separates the values runs across the columns.

This object is 4 columns wide. A matrix carries no per-column schema, so its width is what the catalogue records and there is nothing to list one column at a time. What it holds is described whole instead, summarised over every cell: see Values and Margins above.

120 other datasets carry the same structure fingerprint. That fingerprint is a digest of the column names and their types, and a matrix carries neither, so on one it comes down to the width and the type held: a match is another object of 4 columns holding the same type rather than one with the same columns. The datasets are not listed, because that is not a list of anything this one is related to.