research {GammaTest} | R Documentation |
Calculates the Gamma statistic for a random selection of possible subset of inputs and plots the results in the form of a Gamma histogram.
research(data, percentage=10, plot=TRUE, ...)
data |
Either an input-output dataset, where the outputs are in the last column. |
percentage |
The random sample to take as a percentage of the total number of embeddings. Defaults to 10. |
plot |
Logical indicating whether the Gamma histogram is plotted. Defaults to TRUE. |
... |
Additional Gamma.Test parameters. Use if you wish to change the default values
of p and eps when computing the Gamma Tests. |
Given an input dimension d, there are
2^d - 1
possible input subsets. A full embedding search calculates the Gamma statistic for each subset. The input subsets are then ordered in increasing Gamma value. The random embedding search is used to identify relevent variables for a smooth model when the number of candidate inputs is large i.e. around 20.
Gammas |
An array of Gamma statistics ordered in increasing value. |
mask.array |
An array of ordered masks by Gamma value. |
Samuel E. Kemp. To report any bugs or suggestions please email: sekemp@glam.ac.uk
For papers and Gamma test related material visit http://users.cs.cf.ac.uk:81/Antonia.J.Jones/GammaArchive/IndexPage.htm
# Example on an AR(1) process ar1 <- arima.sim(500, model=list(ar=0.9), sd=sqrt(1)) dvar1 <- dvec(ar1, 10) my.fe <- research(dvar1)