slv.measurements.mobile.location.location_features#

slv.measurements.mobile.location.location_features(gps, cr1000=None, freq='1min', storage_polygons=None, num_processes=1)[source]#

Per-freq GPS features that the classifier needs (plus power, if available).

Columns: n_gps, x/y (UTM median), scatter (max of x/y std, m), speed (median m/s), speed_max, speed_est (position-derived, see MOVING_SPEED_EST), d_track (median distance to any TRAX track, m), d_yard (median distance to the nearest storage polygon, 0 inside), yard_name (that polygon’s name where d_yard ≤ NEAR_YARD), in_yard (fraction of fixes inside a polygon), in_depot (median position inside a load_depot_footprint() shed), nsat (median, if present), and from cr1000: volt (median), volt_min, amb_T, amb_RH. storage_polygons defaults to load_storage_polygons() (all yards); pass a frame with a name column to restrict or replace them. num_processes threads measure the distance of every fix to the track and to the yards (the slow part of a year of fixes; the result is the same for any number; "max" is every CPU the job was given).

Return type:: DataFrame