Turn observed history into a model's uncertainty

Every chosen column of a data frame becomes an empirical() random variable named after the column, and the number of rows becomes the model's scenario count. All columns are read at the same scenario index, so rows observed jointly stay jointly distributed — correlation in the data survives into the model.

Usage

set_empirical(m, data, cols = NULL)

Arguments

m

A model().

data

A data frame of jointly observed rows.

cols

Which columns to use (default: all of them).

Value

The model, invisibly. The handles are retrievable as m$<column>.

Details

A non-numeric column is an error, not a skip: a silently dropped column would leave a model that solves fine and answers the wrong question. Select with cols when the frame carries more than its uncertainty.