Methods_SSI: SSI methods

In MarcooLopez/SFSI_data: Sparse Family and Selection Index

Description

• 'coef': Retrieves the regression coefficients for each value of lambda for each individual in the testing set. If df is specified, only the coefficients for the lambda associated to df are returned. If tst is specified, only the coefficients for the testing individuals given in tst are returned.

• 'fitted': Returns the predicted values for each value of lambda (in columns) for each individual in the testing set (in rows). When using 'lars2' or 'solveEN' functions, a matrix X of predictors is needed

• 'summary': Returns accuracy, MSE, df, achieved by each value of the vector lambda. Also returns summary for the SSI with maximum accuracy and with minimum MSE.

• 'plot': Creates a plot of either accuracy or MSE versus the (average) number of predictors and versus lambda (in negative logarithm).

Usage

## S3 method for class 'SSI'
coef(object, ..., df=NULL, tst=NULL)

## S3 method for class 'SSI'
fitted(object, ...)

## S3 method for class 'SSI'
summary(object, ...)
  
## S3 method for class 'SSI'
plot(..., title=NULL, py=c("accuracy","MSE"))

Arguments

object	An object of the class 'SSI'. One or more objects must be passed as ... in the function plot
df	Average (across testing individuals) number of non-zero regression coefficients
tst	Vector of integers indicating which individuals are in testing set. Default tst=NULL will consider all individuals in object$tst
py	Indicates whether to plot correlation (between observed and predicted) or MSE in y-axis
title	Title of the plot
...	Arguments to be passed: object: One or more objects of the class 'SSI' (for function plot) Not needed for summary and fitted functions

Author(s)

Marco Lopez-Cruz ([e-mail]) and Gustavo de los Campos

Examples

  require(SFSI)
  data(wheatHTP)
  X = scale(X[1:300,])        # Subset and scale markers
  G = tcrossprod(X)/ncol(X)   # Genomic relationship matrix
  y = scale(Y[1:300,"YLD"])   # Subset response variable
  
  fm1 = SSI(y,K=G,tst=1:100,trn=101:length(y),alpha=0.5)
  
  yHat = fitted(fm1)        # Predicted values for each SSI
  corTST = summary(fm1)$accuracy  # Testing set accuracy (correlation cor(y,yHat))
  summary(fm1)$optCOR       # SSI with maximum accuracy
  summary(fm1)$optMSE       # SSI with minimum MSE
  plot(fm1,title=expression('corr('*y[obs]*','*y[pred]*') vs sparsity'))   
  plot(fm1,py="MSE",title='Mean Square Error vs sparsity')     

MarcooLopez/SFSI_data documentation built on April 15, 2021, 10:53 a.m.