Experiment with ChatGPT to generate R code from natural language

rgpt

describe_data_df <- function(x) {
  res_1 <- paste0("data is a dataframe with ", ncol(x), " colums", ":")

  res_2 <- sapply(x, class)
  res_2 <- sapply(1:length(res_2), function(x) {
    paste0("- Column ", x, " is named '", names(res_2)[x], "' and is of class ", res_2[x], ".")
  })
  res_2 <- paste(res_2, collapse = "\n")
  res <- paste(res_1, res_2, sep = "\n")

  return(res)
}


gpt_do <- function(data, prompt) {

  Sys.setenv(OPENAI_VERBOSE = FALSE)
  Sys.setenv(OPENAI_MODEL = "text-davinci-003")
  Sys.setenv(OPENAI_TEMPERATURE = 0)

  if(!is.data.frame(data)) {
    stop("data is not a dataframe.")
  }

  gpt_prompt <- paste0(describe_data_df(data),
                       prompt,
                       "Put the result in an object called 'res'",
                       "Return only the code, do not comment or explain.",
                       sep = "\n")

  gpt_code <- chatgpt::ask_chatgpt(gpt_prompt)
  cat(gpt_code)

  eval(parse(text = gpt_code))
  return(res)
}


# DEMO
library(tidyverse)

# Set your OpenAI API key here
Sys.setenv(OPENAI_API_KEY = "XXXXXXXXXXXXXXXXXXXXXXXXXXXX")

palmerpenguins::penguins %>%
  gpt_do("Use dplyr to add a new variable bill_product by multiplying bill_length_mm by bill_depth_mm") %>%
  filter(!is.na(sex)) %>%
  gpt_do("Make a scatterplot with ggplot showing the relationship between bill_product and body_mass_g.
          Create a panel for each sex. Use a bright theme. Rewrite the axis labels")