estimate_program: Estimate a program (R function) from data.
In JohnNay/agp: Agent Genetic Programming

estimate_program uses rgp's rgp::typedGeneticProgramming.

estimate_program(formula, data, subset = NULL, loss = c("log_lik", "rmse",
  "identity", "identity_multi_class"), identity_outcome_type = c("integer",
  "character", "factor"), link = c("logit", "probit", "cauchit", "identity"),
  func_list = list("+", "-", "*", "divide", "exp", ">", "<", "logn", "sqrtn",
  "&", "|", "!", "ifelse2", "one_rnorm"), mins = 10, steps = 2000,
  repeats = 2, parallel = FALSE, cores = NULL,
  enable_complexity = FALSE, lambda = 50, crossover_probability = 0.5)

`formula`	formula used to create the data.frame needed, ensuring that the outcome variable, the variable to the left of "~", is the first column in the data.frame and the following columns are the predictor variables.
`data`	a data.frame with named columns containing the variables in formula. Neither a matrix nor an array will be accepted. We use the `formula` to turn this into a data.frame where (i.) the first column is named `"outcome"` and has the outcome variable we are evolving programs to predict; (ii.) all other columns (there must be at least one other column) that follow the `"outcome"` column are named columns containing the predictor variables, which can be of any type. The order matters: when using the evolved program for predictions the (named) arguments to the function will be the predictor variables in the order they are supplied to this `estimate_program()` function.
`subset`	a specification of the rows to be used: defaults to all rows. This can be any valid indexing vector (see [.data.frame) for the rows of data or if that is not supplied, a data frame made up of the variables used in formula.
`loss`	Optional Character vector length one
`identity_outcome_type`	Optional Character vector length one only needed if `loss=="identity"`.
`link`	Optional Character vector length one
`func_list`	Optional List where each element is a length one character vector.
`mins`	Optional Integer vector length one
`steps`	Optional Integer vector length one
`repeats`	Optional Integer vector length one sepcifying how many times to repeat the model fitting
`parallel`	Optional Logical vector length one. Default is `parallel = FALSE`; `parallel = TRUE` can be slower if the data set is small relative to the numner of population evolutions desired
`cores`	Optional Integer vector length one
`enable_complexity`	Optional logical vector length one
`lambda`	Optional integer vector length one for the number of children rgp::typedGeneticProgramming() creates in each generation
`crossover_probability`	Optional numeric vector length one (default == 0.5) for rgp::typedGeneticProgramming()