Validate surveillance data — check

Validate and standardise surveillance data for use throughout the package. The returned incast_data object can be passed directly to forecasting and nowcasting functions.

Usage

check_data(data, key = "location")

Arguments

data: A data frame containing target_end_date (Date), observation (numeric), target (character), and one or more key columns. An optional as_of (Date) column enables nowcasting with get_ncast. If data is already an incast_data object, it is returned unchanged.
key: Character vector giving the column name(s) that uniquely identify each time series, equivalent to the key of a tsibble. Each unique combination of key values is treated as a separate series. Defaults to "location".

Value

An incast_data object containing:

data: Validated data with standardised column types.
key: Names of the key columns.
target: Target variable name.
window: Start and end dates of the data.
interval: Reporting interval in days (for example, 7 for weekly data).
history: Logical indicating whether revision history (as_of) is available.

Details

Data must contain one row per time series and reporting date (and as_of, if present). All series must have the same reporting interval, share the same reporting dates, and end on the same date. Series may begin at different times and may contain missing reporting periods.

Author

Cyril Geismar

Examples

if (FALSE) { # \dontrun{
x <- get_data("covid", c("ny", "ca")) |> check_data()
my_x <- read.csv("my_data.csv") |>
  check_data(key = c("location", "age_group"))
} # }