PP_Index: Function to find the Projection Pursuit indexes (PP).

Description

Function used to find Projection Pursuit indexes (PP).

Usage

PP_Index(data, class = NA, vector.proj = NA, 
         findex = "HOLES", dimproj = 2, weight = TRUE, 
         lambda = 0.1, r = 1, ck = NA)

Value

num.class: Number of classes.
class.names: Class names.
findex: Projection index function used.
vector.proj: Projection vectors found.
index: Projection index found in the process.

Arguments

data: Numeric dataset without class information.
class: Vector with names of data classes.
vector.proj: Vector projection.
findex: Projection index function to be used:
"lda" - LDA index,
"pda" - PDA index,
"lr" - Lr index,
"holes" - Holes index (default),
"cm" - Central Mass index,
"pca" - PCA index,
"friedmantukey" - Friedman Tukey index,
"entropy" - Entropy index,
"legendre" - Legendre index,
"laguerrefourier" - Laguerre Fourier index,
"hermite" - Hermite index,
"naturalhermite" - Natural Hermite index,
"kurtosismax" - Maximum kurtosis index,
"kurtosismin" - Minimum kurtosis index,
"moment" - Moment index,
"mf" - MF index,
"chi" - Chi-square index.
dimproj: Dimension of data projection (default = 2).
weight: Used in index LDA, PDA and Lr to weight the calculations for the number of elements in each class (default = TRUE).
lambda: Used in the PDA index (default = 0.1).
r: Used in the Lr index (default = 1).
ck: Internal use of the CHI index function.

Author

Paulo Cesar Ossani

Marcelo Angelo Cirillo

References

Ossani, P. C.; Figueira, M. R.; Cirillo, M. A. Proposition of a new index for projection pursuit in the multiple factor analysis. Computational and Mathematical Methods, v. 1, p. 1-18, 2020.

Cook, D.; Buja, A.; Cabrera, J. Projection pursuit indexes based on orthonormal function expansions. Journal of Computational and Graphical Statistics, 2(3):225-250, 1993.

Cook, D.; Buja, A.; Cabrera, J.; Hurley, C. Grand tour and projection pursuit, Journal of Computational and Graphical Statistics, 4(3), 155-172, 1995.

Cook, D.; Swayne, D. F. Interactive and Dynamic Graphics for data Analysis: With R and GGobi. Springer. 2007.

Espezua, S.; Villanueva, E.; Maciel, C. D.; Carvalho, A. A projection pursuit framework for supervised dimension reduction of high dimensional small sample datasets. Neurocomputing, 149, 767-776, 2015.

Friedman, J. H., Tukey, J. W. A projection pursuit algorithm for exploratory data analysis. IEEE Transaction on Computers, 23(9):881-890, 1974.

Hastie, T., Buja, A., Tibshirani, R. Penalized discriminant analysis. The Annals of Statistics. 23(1), 73-102 . 1995.

Huber, P. J. Projection pursuit. Annals of Statistics, 13(2):435-475, 1985.

Jones, M. C.; Sibson, R. What is projection pursuit, (with discussion), Journal of the Royal Statistical Society, Series A 150, 1-36, 1987.

Lee, E. K.; Cook, D. A projection pursuit index for large p small n data. Statistics and Computing, 20(3):381-392, 2010.

Lee, E.; Cook, D.; Klinke, S.; Lumley, T. Projection pursuit for exploratory supervised classification. Journal of Computational and Graphical Statistics, 14(4):831-846, 2005.

Martinez, W. L., Martinez, A. R.; Computational Statistics Handbook with MATLAB, 2th. ed. New York: Chapman & Hall/CRC, 2007. 794 p.

Martinez, W. L.; Martinez, A. R.; Solka, J. Exploratory data Analysis with MATLAB, 2th. ed. New York: Chapman & Hall/CRC, 2010. 499 p.

Pena, D.; Prieto, F. Cluster identification using projections. Journal of the American Statistical Association, 96(456):1433-1445, 2001.

Posse, C. Projection pursuit exploratory data analysis, Computational Statistics and data Analysis, 29:669-687, 1995a.

Posse, C. Tools for two-dimensional exploratory projection pursuit, Journal of Computational and Graphical Statistics, 4:83-100, 1995b.

Examples

Run this code

data(iris) # data set

data <- iris[,1:4]

# Example 1 - Without the classes in the data
ind <- PP_Index(data = data, class = NA, vector.proj = NA, 
                findex = "moment", dimproj = 2, weight = TRUE, 
                lambda = 0.1, r = 1)

print("Number of classes:"); ind$num.class
print("class Names:"); ind$class.names
print("Projection index function:"); ind$findex
print("Projection vectors:"); ind$vector.proj  
print("Projection index:"); ind$index


# Example 2 - With the classes in the data
class <- iris[,5] # data class

findex <- "pda" # index function

sphere <- TRUE # spherical data

res <- PP_Optimizer(data = data, class = class, findex = findex,
                    optmethod = "SA", dimproj = 2, sphere = sphere, 
                    weight = TRUE, lambda = 0.1, r = 1, cooling = 0.9, 
                    eps = 1e-3, maxiter = 1000, half = 30)

# Comparing the result obtained
if (match(toupper(findex),c("LDA", "PDA", "LR"), nomatch = 0) > 0) {
  if (sphere) {
     data <- apply(predict(prcomp(data)), 2, scale) # spherical data
  }
} else data <- as.matrix(res$proj.data[,1:Dim])

ind <- PP_Index(data = data, class = class, vector.proj = res$vector.opt, 
                findex = findex, dimproj = 2, weight = TRUE, lambda = 0.1,
                r = 1)

print("Number of classes:"); ind$num.class
print("class Names:"); ind$class.names
print("Projection index function:"); ind$findex
print("Projection vectors:"); ind$vector.proj  
print("Projection index:"); ind$index
print("Optimized Projection index:"); res$index[length(res$index)]

Run the code above in your browser using DataLab

Last chance! 50% off unlimited learning