get.q {EMMAgeo}

Generate a parameter matrix with q.min and q.max values for robust EMMA.

Description

This function uses the input data matrix X and a vector of weight transformation limits to generate a matrix of minimum and maximum likely numbers of end-members to be used to model and extract robust end-members.

Usage

get.q(
  X,
  l = 0,
  q.min = 2,
  q.max = 10,
  criteria.min = 0.5,
  criteria.max = "local_max",
  correct.output = TRUE,
  ...
)

Arguments

X

Numeric matrix, input data set with m samples (rows) and n variables (columns).

l

Numeric vector, weight transformation limits, default is zero.

q.min

Numeric scalar, minimum number of end-members to use, default is 2.

q.max

Numeric scalar, maximum number of end-members to use, default is 10.

criteria.min

Numeric scalar, minimum value of explained variance reached to be a valid model realisation, default is 0.5.

criteria.max

Character or numeric scalar, either keyword "local_max" to use first local maximum or any numeric value of explained variance, default is "local_max".

correct.output

Logical scalar, option to correct the output for twisted values and remove combinations with NA-values. See details.

...

Further arguments, passed to the function.

Details

The parameter q.min should be at least 2 because otherwise the entire dataset would consist of one end-member and there would be no variability at all. The parameter q.max is set to 10 by default, based on practical issues. In natural systems, there are only rarely occasions when such a high number of sediment transport regimes may be preserved in and can be resolved from sedimentary deposits. The parameter l should be a vector between the minimum possible (zero) and maximum possible value (by definition the median, 0.5, but usually a lower value). Whensubmitting only a scalar, the variability can be only due to the range of possible endmembers (between q.min and q.max). If the parameter correct.output is enabled, this can decrease the number of valid values for l, i.e. the number of rows of the output matrix may no longer be the same as the length of the input vector of l. In such a case the vector l must be replaced by the rownames of the output matrix (l <- as.numeric(rownames(get.q()))).

Value

Numeric matrix of class "EMMAgeo_q", minimum and maximum numbers of end-members as well as corresponding weight transformation values as rownames.

Author(s)

Michael Dietze, Elisabeth Dietze

References

Dietze E, Hartmann K, Diekmann B, IJmker J, Lehmkuhl F, Opitz S, Stauch G, Wuennemann B, Borchers A. 2012. An end-member algorithm for deciphering modern detrital processes from lake sediments of Lake Donggi Cona, NE Tibetan Plateau, China. Sedimentary Geology 243-244: 169-180.

See Also

EMMA, test.parameters, test.robustness

Examples

## load example data set
data("example_X")

## create parameter matrix
get.q(X = X, l = c(0, 0.05, 0.10, 0.15))