bmopPar: Set Rbmop parameters

In gherardovarando/Rbmop: B-spline MoPs densities and conditional densities

Description

Set appropriate global parameters to be used by Rbmop functions . This is the appropriate way to change those parameters. The use is similar to par() in package graphics.

Usage

bmopPar(...)

Arguments

...	see details.

Details

This function set parameters used for estimation.The call bmopPar() just print the values of the paramters.

mle=FALSE: logical. If use maximum-likelihood estimation of bmop coefficient, setting mle=TRUE just force repMax=1.

N=NA: If present, the number of knots in every dimensions. Vector of positive integer, if needed values will be recycled.

order=3: The order of the B-spline in every dimensions, vector of positive integer, if needed values will be recycled.

alpha=3: The penalization exponent to compute the number of knots. This is the default method to compute the number of knots, with the formula: floor(n^(1/alpha)), where n is the number of observations in the dataset. If !is.na(N) then the number of knots will be set to N^d where d is the number of dimensions in the dataset (num. of variables).

knotsMethod="uniform": "uniform" or "quantiles" knots, how knots are computed by generate_knots.

k=2: Coefficient of AIC (penalized likelihood), positive integer or "BIC" string. This is used by search_bmop.

toll=10^{-10}: Tollerance for the increment of the likelihood in the mle estimation of the coefficient.

repMax=100: maximum number of iteration in the mle estimation of the coefficient

MIN=10^{-10}: This is not a learning parameter but instead define the MIN parameter in the evaluation of bmop object. Observe that some functions like logLik or plot, set this parameter independently.

autoReduce=200: This value set the maximum dimension of an accepted dataset as raw-data, for larger dataset, functions bmop_fit and search_bmop will be applied over the reduced bins (histogram). Setting it to Inf disable this features.

warnings=TRUE: boolean, setting this value to FALSE, avoid the warnings generated by autoreduce.

gherardovarando/Rbmop documentation built on May 17, 2019, 4:17 a.m.