-
Notifications
You must be signed in to change notification settings - Fork 16
Open
Description
Line 75 in e2b34e9
| self._coverage = np.max(np.min(self._dist_land_data, 1)) |
In
self._coverage = np.max(np.min(self._dist_land_data, 1))
this value is always 0 for a proper distance matrix (dist_land_data is a subset of the full distance matrix, but for each landmark there is one data point (namely itself), to which it has distance 0.
Or do we rather mean np.min(self._dist_land_data, 0) which is the distance of each data point to the closest landmark (which will evaluate 0 for the data points which are also chosen as landmarks) but otherwise a non-negative value.
np.max(np.min(self._dist_land_data, 0)) is then the largest distance of a datapoint-landmark pair.
Reactions are currently unavailable
Metadata
Metadata
Assignees
Labels
No labels