Gap-Fill a Single Region-Indicator Time Series — gapfill

Interpolates annual values between DHS survey observations for a single region-indicator series. Uses FMM cubic spline for point estimates and penalized GAM for calibrated uncertainty intervals.

Usage

gapfill_series(years, values, target_years = NULL,
  transform = c("log", "logit"), level = 0.95,
  sigma_floor = 0.25, forecast_to = NULL)

Arguments

years: Integer vector of survey years (the observed time points).
values: Numeric vector of observed indicator values.
target_years: Integer vector of years at which to predict. Defaults to annual sequence from min to max observed year.
transform: Character: "log" for rates or "logit" for proportions.
level: Numeric confidence level for prediction intervals (default: 0.95).
sigma_floor: Numeric minimum prediction SE (default: 0.25).
forecast_to: Integer: if set, forecasts forward to this year using ETS.

Value

A tibble with columns: year, estimate, ci_lo, ci_hi, source.

Examples

if (FALSE) { # \dontrun{
gf <- gapfill_series(c(2005, 2010, 2015), c(90, 85, 80),
  transform = "log")
} # }