Cross validation for estimating the classification rate of a discrminant analysis for directional data assuming a Purkayastha distribution.
purka.da(y, ina, fraction = 0.2, R = 100, seed = NULL)A list including:
The estimated percent of correct classification and two estimated standard deviations. The one is the standard devation of the rates and the other is assuming a binomial distribution.
Three types of confidence intervals, the standard one, another one based on the binomial distribution and the third one is the empirical one, which calcualtes the upper and lower 2.5% of the rates.
A numerical vector with data expressed in radians, or a matrix with two columns (cos and sin) for circular data. Or a matrix with 3 columns (unit vectors) for spherical data.
A variable indicating the groupings.
The fraction of data to be used as test set.
The number of repetitions.
You can specify your own seed number here or leave it NULL.
Michail Tsagris.
R implementation and documentation: Michail Tsagris mtsagris@uoc.gr.
A repeated cross validation procedure is performed to estimate the rate of correct classification.
Purkayastha S. (1991). A Rotationally Symmetric Directional Distribution: Obtained through Maximum Likelihood Characterization. The Indian Journal of Statistics, Series A, 53(1): 70-83
Cabrera J. and Watson G. S. (1990). On a spherical median related distribution. Communications in Statistics-Theory and Methods, 19(6): 1973-1986.
vmf.da, vmfda.pred, dirknn
x <- rvmf(100, rnorm(3), 15)
ina <- rep(1:2, each = 50)
purka.da(x, ina, fraction = 0.2, R = 50)
Run the code above in your browser using DataLab