Declustered statistics
Declustered statistics are statistics that make use of geospatial coordinates to correct potential sampling bias:
We provide the following declustered statistics:
Statistics.mean — Method
mean(geotable, column)
mean(geotable, column, side)Declustered mean of given column in geotable. Optionally, specify the declustering block side.
Statistics.var — Method
var(data, column)
var(data, column, side)Declustered variance of given column in geotable. Optionally, specify the declustering block side.
Statistics.quantile — Method
quantile(data, column, prob)
quantile(data, column, prob, side)Declustered quantile of given column in geotable at given probability. Optionally, specify the declustering block side.
GeoStatsBase.histogram — Method
histogram(geotable, column; kwargs...)
histogram(geotable, column, side; kwargs...)Declustered histogram of given column in geotable. Optionally, specify the declustering block side and forward keyword arguments kwargs to StatsBase.fit(Histogram, ...).
The following example shows bias towards high Au values:
gtb = GeoIO.load("data/clustered.csv", coords = ("x", "y"))
44 rows omitted
| 54×2 GeoTable over 54 PointSet | |
| Au | geometry |
|---|---|
| Continuous | Point |
| [NoUnits] | 🖈 Cartesian{NoDatum} |
| 0.280863 | (x: 10.0 m, y: 230.0 m) |
| 0.145002 | (x: 10.0 m, y: 250.0 m) |
| 0.991987 | (x: 30.0 m, y: 50.0 m) |
| 0.117489 | (x: 30.0 m, y: 70.0 m) |
| 0.672637 | (x: 30.0 m, y: 130.0 m) |
| 0.270452 | (x: 50.0 m, y: 10.0 m) |
| 0.380441 | (x: 50.0 m, y: 30.0 m) |
| 0.10795 | (x: 50.0 m, y: 70.0 m) |
| 0.108283 | (x: 50.0 m, y: 170.0 m) |
| 0.606969 | (x: 50.0 m, y: 190.0 m) |
| ⋮ | ⋮ |
mean(gtb."Au")0.5639250835599507mean(gtb, "Au")0.4771220767306614quantile(gtb."Au", 0.5)0.597289708842421quantile(gtb, "Au", 0.5)0.4713802469107024