gx.runs: The Wald-Wolfowitz, 'Runs', Test

In rgr: Applied Geochemistry EDA

Description

The ‘runs’ test is used to infer whether two states, e.g., > and < some threshold are mutually independent along a traverse. In applied geochemical terms, it tests for pattern coherence. If the pattern of runs is not coherent at the scale of the sampling it will be difficult to identify any spatially consistent dispersion processes.

Usage

gx.runs(n1, n2, u)

Arguments

n1	the number of < threshold sites along a traverse.
n2	the number of > threshold sites along a traverse.
u	the number of runs of > and < threshold sites along the traverse.

Note

Given a priori information on the location of a mineral occurrence, the gx.hypergeom function provides a far more insightful test. The ‘runs’ test is better suited for evaluating patterns due to lithological or environmental changes along a traverse when some ‘threshold’ can be selected that differentiates between two patterns

Author(s)

Robert G. Garrett

References

Stanley, C.R., 2003. Statistical evaluation of anomaly recognition performance. Geochemistry: Exploration, Environment, Analaysis, 3(1):3-12.

See Also

gx.hypergeom

Examples

## From Stanley (2003) Table 2

gx.runs(27, 4, 7)
gx.runs(25, 6, 7)
gx.runs(28, 3, 5)

Example output

Loading required package: MASS
Loading required package: fastICA
  Traverse length is 31 sites with 4 anomalous sites in 7 runs
  E(u) = 7.97 , Var(u) = 1.39 and z = -0.822 
  Probability that this is due to 'chance' is 20.6%
  Traverse length is 31 sites with 6 anomalous sites in 7 runs
  E(u) = 10.68 , Var(u) = 2.8 and z = -2.198 
  Probability that this is due to 'chance' is 1.4%
  Traverse length is 31 sites with 3 anomalous sites in 5 runs
  E(u) = 6.42 , Var(u) = 0.8 and z = -1.589 
  Probability that this is due to 'chance' is 5.6%

rgr documentation built on May 2, 2019, 6:09 a.m.