scale_abiotic_for_fit()

scale_abiotic_for_fit

R Documentation

Scale Abiotic Data for Model Fitting

Description

Centres and scales abiotic predictor variables, records the scaling attributes for later back-transformation, and returns both the scaled data frame and the attributes as a named list.

‘age’ is centred only (mean subtracted, no division by SD). All other variables are both centred and scaled (divided by SD) when more than one sample is present.

Usage

scale_abiotic_for_fit(data_abiotic_wide = NULL)

Arguments

data_abiotic_wide

A data frame in wide format as returned by ‘prepare_abiotic_for_fit()’, containing real columns ‘dataset_name’, ‘age’, and one column per abiotic variable.

Details

Rows with any ‘NA’ across the abiotic variables are silently dropped via ‘tidyr::drop_na()’ before scaling. The returned ‘scale_attributes’ list preserves the same structure as ‘attributes(scale(x))[-1]’ (i.e., ‘dim’ excluded).

Value

A named list with two elements:

‘data_abiotic_scaled’: A data frame with row names in the format ’"<dataset_name>__<age>"‘, an ’age’ column (centre-only scaled), and all other abiotic variable columns (centre-and-scale). Rows with any ‘NA’ are dropped before scaling.
‘scale_attributes’: A named list of ‘center’ and ‘scale’ attributes for each variable (including ‘age’), which can be used to back-transform predictions.