Skip to contents

write_spec_yaml() writes one specification; write_specs_yaml() writes a sweep – several specifications in one file, with the settings they share hoisted into a defaults: block so the file shows exactly what differs between them. read_specs() reads either form (or every YAML file in a directory) and read_spec_yaml() reads a single-spec file.

Usage

write_spec_yaml(spec, path)

read_spec_yaml(path)

read_specs(path)

write_specs_yaml(specs, path, title = "sdidlab specifications")

Arguments

spec

An sdid_spec().

path

File to write, or NULL to return the text instead. For read_specs(), a file or a directory.

specs

A list of sdid_spec()s.

title

Comment line at the top of a sweep file.

Value

The writers return the path invisibly (or the YAML text when path = NULL); read_specs() returns a list of specifications and read_spec_yaml() a single one.

Details

Writing then reading is an identity: the specifications that come back have the same spec_hash() as those written.

A sweep file looks like this:

defaults:
  design: {treat_time: 1989}
  inference: {replications: 100}
specs:
- label: SDID
- label: Synthetic control
  estimator: sc

Examples

specs <- list(
  sdid_spec(label = "SDID", design = list(treat_time = 1989)),
  sdid_spec(label = "SC", estimator = "sc", design = list(treat_time = 1989))
)
cat(write_specs_yaml(specs, NULL, title = "Prop 99 estimators"))
#> # Prop 99 estimators
#> # Placebo standard errors depend on the RNG stream: fit these in file
#> # order under one seed, or with run_specs(seed_each = TRUE).
#> defaults:
#>   design:
#>     treat_time: 1989.0
#> specs:
#> - label: SDID
#> - label: SC
#>   estimator: sc

f <- tempfile(fileext = ".yml")
write_specs_yaml(specs, f)
back <- read_specs(f)
identical(vapply(back, spec_hash, ""), vapply(specs, spec_hash, ""))
#> [1] TRUE