AnalyzeTS (version 2.2)

GDOC: Finding the best C values

Description

Finding the best C values for more Abbasov Mamedova and NFTS models according to DOC algorithm at the same time.

Usage

GDOC(ts, n = 7, w = 7, D1 = 0, D2 = 0, error = 1e-06, k = 500, r = 13, 
CEF = "MSE", type = "Abbasov-Mamedova", show.complete = TRUE)

Arguments

ts

Univariate time series.

n

A numeric vector where each element is number of fuzzy set.

w

A numeric vector where each element is w parameter.

D1,D2

Two proper positive numbers.

error

Error of C value is finded by DOC algorithm, which compare the best C value really. Default error = 0.000001.

k

In each iteration of the algorithm, k+1 (or k or k-1) values of C will be considered. The k must be a integer and greater than 499, default k = 500.

r

Display results returned to the specified number of decimal places (default 13). (See round2str for details of r paramicter.)

CEF

One of the criterion to evaluate forecasting model, must be one of "ME","MAE" , "MPE", "MAPE", "MSE" (default), or "RMSE".

type

A character vector where each element is choosing model to predicts time series by fuzziness, type = "Abbasov-Manedova" (default) or type = "NFTS" or both.

show.complete

If TRUE, a graph will appear showing the percentage completed.

Value

A list contain two components where the first component is the best C values of Abbasov-Mamedova models and the second component is the best C values of NFTS models.

Details

GDOC function consider length(n)*length(w)*length(type) models combining from three parameter n, w and type, and then using DOC function finding the best C values forone by models.

See Also

Using DOC function in case only a Abbasov-Mamedova or NFTS model.

Examples

#For examples see example(Gfuzzy.ts2)