Skip to contents

Retrieve data from an IMF dataset

Source: R/imf_get.R
imf_get.Rd

Fetches observations for a given dataflow_id and resource_id from the IMF SDMX 3.0 Data API. The request key is constructed from the dataset's datastructure (DSD) using the positional order of dimensions. Time filtering is applied via query parameters.

Usage

imf_get(
  dataflow_id,
  dimensions = list(),
  start_period = NULL,
  end_period = NULL,
  progress = FALSE,
  max_tries = 10L,
  cache = TRUE
)

Arguments

dataflow_id

Character scalar. The dataflow to query (e.g., "GFS").

dimensions

Named list mapping dimension IDs to character vectors of codes to include. Omitted dimensions are wildcarded in the key. Each dimension position in the DSD corresponds to one dot-separated slot in the key; multiple codes per slot are joined by '+'.

start_period

Optional character. Lower bound for time filtering (e.g., "2000", "2000-Q1", "2000-01").

end_period

Optional character. Upper bound for time filtering, same format as start_period. The request always uses the SDMX 3.0 dataflow context under the hood and sets dimensionAtObservation = "TIME_PERIOD" to request a time-series view.

progress

Logical; whether to show request progress.

max_tries

Integer; maximum retry attempts for HTTP requests.

cache

Logical; whether to enable caching for HTTP requests.

Value

A tibble with one row per observation, including dimension columns, time period, value column(s), and any requested attributes. Exact column names follow the dataset's DSD and may vary by dataflow_id.

Details

By default, the request targets the all agencies scope for the data path, assuming dataflow IDs are globally unique in practice. The response layout uses a time-series context, and client code will shape the parsed payload into a tidy tibble.

The request key is built by ordering dimensions by their DSD position and filling each position with either a '+'-joined set of selected codes or a blank for wildcard. Time filtering is applied via start_period and end_period query parameters rather than encoding time into the key.

Examples

# \donttest{
if (curl::has_internet()) {
  imf_get(
    dataflow_id = "FM",  # Fiscal Monitor
    dimensions = list(COUNTRY = c("USA", "CAN"))
  )
}
#> # A tibble: 552 × 5
#>    COUNTRY INDICATOR         FREQUENCY TIME_PERIOD OBS_VALUE
#>    <chr>   <chr>             <chr>     <chr>           <dbl>
#>  1 CAN     CAB_S13_POPGDP_PT A         1991           -7.61 
#>  2 CAN     CAB_S13_POPGDP_PT A         1992           -8.07 
#>  3 CAN     CAB_S13_POPGDP_PT A         1993           -8.05 
#>  4 CAN     CAB_S13_POPGDP_PT A         1994           -7.08 
#>  5 CAN     CAB_S13_POPGDP_PT A         1995           -5.58 
#>  6 CAN     CAB_S13_POPGDP_PT A         1996           -2.44 
#>  7 CAN     CAB_S13_POPGDP_PT A         1997            0.315
#>  8 CAN     CAB_S13_POPGDP_PT A         1998            0.346
#>  9 CAN     CAB_S13_POPGDP_PT A         1999            1.29 
#> 10 CAN     CAB_S13_POPGDP_PT A         2000            1.70 
#> # ℹ 542 more rows
# }