Suppose that you want to write a code to perform some task such as estimating a model or simulation data.

Steps

In the tests/ folder, create a file to develop procedures, say, tests/test_estimate_model.R.
In this file, first write a procedure to be performed. git commit once this is done.

# expand asic spec vector for top currencies
algo_date <-
  spec %>%
  tidyr::expand(
    tidyr::nesting(model, algo),
    date = seq(min(release), to = as.Date("2020-12-31"), by = 1)
  ) %>%
  dplyr::left_join(
    spec,
    by = c("model", "algo", "date" = "release")
  ) %>%
  dplyr::group_by(model) %>%
  tidyr::fill(dplyr::everything(), .direction = "down") %>%
  dplyr::ungroup()
# make baseline data
# complete in asic x (currency x algo x date)
# include NA
base_currency_algo_date <-
  currency_algo_date %>%
  dplyr::filter(asic_dummy)
base_currency_algo_date <-
  base_currency_algo_date %>%
  tidyr::expand(
    model = unique(algo_date$model),
    tidyr::nesting(currency, algo, date)
  ) %>%
  dplyr::left_join(
    algo_date,
    by = c("model", "algo", "date")
  ) %>%
  dplyr::filter(!is.na(hash_rate_spec)) %>%
  dplyr::left_join(
    currency_algo_date,
    by = c("currency", "algo", "date")
  ) %>%
  dplyr::left_join(
    spec %>%
      dplyr::select(model, release),
    by = "model"
  )
base_currency_algo_date <-
  base_currency_algo_date %>%
  dplyr::mutate(currency_base = (currency == currency_base))

Then, turn it into a function. git commit once this is done.

# make baseline data for computing utility and choice probability
make_baseline_data_asic <-
  function(
		spec, 
		currency_algo_date, 
		currency_base
		) {
	    # expand asic spec vector for top currencies
	    algo_date <-
	      spec %>%
	      tidyr::expand(
	        tidyr::nesting(model, algo),
	        date = seq(min(release), to = as.Date("2020-12-31"), by = 1)
	      ) %>%
	      dplyr::left_join(
	        spec,
	        by = c("model", "algo", "date" = "release")
	      ) %>%
	      dplyr::group_by(model) %>%
	      tidyr::fill(dplyr::everything(), .direction = "down") %>%
	      dplyr::ungroup()
	    # make baseline data
	    # complete in asic x (currency x algo x date)
	    # include NA
	    base_currency_algo_date <-
	      currency_algo_date %>%
	      dplyr::filter(asic_dummy)
	    base_currency_algo_date <-
	      base_currency_algo_date %>%
	      tidyr::expand(
	        model = unique(algo_date$model),
	        tidyr::nesting(currency, algo, date)
	      ) %>%
	      dplyr::left_join(
	        algo_date,
	        by = c("model", "algo", "date")
	      ) %>%
	      dplyr::filter(!is.na(hash_rate_spec)) %>%
	      dplyr::left_join(
	        currency_algo_date,
	        by = c("currency", "algo", "date")
	      ) %>%
	      dplyr::left_join(
	        spec %>%
	          dplyr::select(model, release),
	        by = "model"
	      )
	    base_currency_algo_date <-
	      base_currency_algo_date %>%
	      dplyr::mutate(currency_base = (currency == currency_base))
	    return(base_currency_algo_date)
  }

Cut this function and paste it to a file in R/ folder that collects R function definitions, say, R/functions.R.
devtools::load_all(".") to load the function definition as a method in the project package.
Check that there is no global reference in the function by running codetools::findGlobals(make_baseline_data_asic) in the R console. This will list all global variables used in the function. Check there are no user-defined global variables in the list. The variable names used in dplyr is recognized as global variable, but they are okay. git commit once this is done.

> codetools::findGlobals(make_baseline_data_asic)
 [1] "!"              "$"              "%>%"            "("              "::"             "{"             
 [7] "<-"             "=="             "algo"           "as.Date"        "asic_dummy"     "c"             
[13] "currency"       "date"           "hash_rate_spec" "is.na"          "min"            "model"         
[19] "release"        "return"         "seq"            "unique"

In tests/test_estimate_model.R, keep a line to perform the function:

# make basic data
base_currency_algo_date <- 
  make_baseline_data_asic(
    spec = spec, 
    currency_algo_date = currency_algo_date,
    currency_base = currency_base
  )

Continue this procedure until the task is completed.git commit once this is done. It should be like this:

test_estimate_model

As the task is completed in tests/test_estimate_model.R, make a file in the main/ folder to run the procedure while cutting unnecessary steps, say, main/estimate_model.R . git commit once this is done. It should look like this:

estimate_model

If you want to add a new procedure in this task or would like to modify existing functions, you first try it in tests/test_estimate_model.R. You can work on main/estimate_model.R only after you confirm the performance of the procedure inside tests/test_estimate_model.R.

Steps

Rationale