Run a 3-d (lat, lon, time) Kulldorff scan with MC inference

Usage

mrm_tps_kulldorff_scan(
  data,
  date_col = "OCC_DATE",
  lat_col = "LAT_WGS84",
  lon_col = "LONG_WGS84",
  radii_km = c(1, 2, 3, 5, 8),
  window_years = 4,
  n_centers = 60L,
  n_permutations = 199L,
  n_top_clusters = 1L,
  seed = 42L
)

Arguments

data: data.frame with date_col, lat_col, lon_col.
date_col: Column name of the event date (default "OCC_DATE").
lat_col, lon_col: WGS84 lat/long column names.
radii_km: Candidate cylinder radii in km.
window_years: Time-cylinder length in years.
n_centers: Number of random candidate centres sub-sampled.
n_permutations: Monte-Carlo permutations.
n_top_clusters: Integer; number of top clusters to return. Accepted for Python signature parity. The current implementation returns a single primary cluster (the secondary-cluster loop in morie.mrm_kulldorff.py breaks out pending a proper mask-and-rescan rewrite); values >1 are reserved for that future TRUE multi-cluster mode.
seed: Random seed.

Value

A one-row data.frame describing the top cluster, with columns center_lat, center_lon, radius_km, t_start, t_end, n_observed, n_expected, relative_risk, log_lrt, p_value.

Examples

if (FALSE) {
  tps <- morie_sample("tps_assault")
  mrm_tps_kulldorff_scan(tps, n_permutations = 49)
}