Skip to main content

Table 1 Bayesian Information Criteria (BIC) for normal and t mixture models.

From: Modeling the probability distribution of positional errors incurred by residential address geocoding

Error dataset Distribution Number of Components BIC
(a) Normal 1 48103
  Normal 2 45851
  Normal 3 45236
  Normal 4 45124
  t 1 46083
  t 2 45358
  t 3 45056
  t 4 45042
(b) Normal 1 46422
  Normal 2 44809
  Normal 3 44597
  Normal 4 44557
  t 1 45659
  t 2 44538
  t 3 44516
  t 4 44459
(c) Normal 1 67174
  Normal 2 63174
  Normal 3 62710
  Normal 4 62446
  t 1 62841
  t 2 62345
  t 3 62219
  t 4 62230
(d) Normal 1 64227
  Normal 2 61360
  Normal 3 61101
  Normal 4 61059
  t 1 61092
  t 2 60980
  t 3 60982
  t 4 60994
  1. Models with several different numbers of components, were fitted to the following four error datasets: (a) automated geocoding positional errors; (b) automated geocoding positional errors aligned with axial direction of corresponding street segment; (c) E911 positional errors; (d) E911 positional errors aligned with axial direction of corresponding street segment.