parcorr.pvalue: Partial correlation, p-value and t-statistic associated with...
In rakheon/d2wlasso: Structured variable selection with weighted lasso

Description Usage Arguments Value Examples

Partial correlation, p-value and t-statistic associated with the regression between response y and a covariate x after adjustment for a covariate z.

parcorr.pvalue(
  factor.z,
  x,
  y,
  z,
  delta = NULL,
  method = "pearson",
  alternative = "two.sided",
  ttest.pvalue = FALSE,
  regression.type
)

`factor.z`	logical. If TRUE, the fixed variable z is a factor variable.
`x`	(n by 1) matrix corresponding to a covariate vector where n is the sample size.
`y`	(n by 1) a matrix corresponding to the response variable. If `regression.type` is "cox", `y` contains the observed event times.
`z`	(n by 1) matrix of additional fixed covariate affecting response variable. Can be NULL.
`delta`	(n by 1) a matrix that denotes censoring when `regression.type` is "cox" (1 denotes survival event is observed, 0 denotes the survival event is censored). Can be NULL.
`method`	character indicating the type of correlation to compute. Default if "pearson"
`alternative`	character indicating whether the p-value is computed using one-sided or two-sided testing. Default is "two-sided".
`ttest.pvalue`	logical indicator. If TRUE, p-value for each covariate is computed from univariate linear/cox regression of the response on each covariate. If FALSE, the p-value is computed from correlation coefficients between the response and each covariate. Default is FALSE.
`regression.type`	a character indicator that is either "linear" for linear regression or "cox" for Cox proportional hazards regression. Default is "linear".

p.value:p-value of the coefficient of x in the regression of y on x and z.
estimate:Partial correlation between x and y after adjustment for z.
t.stat:t-statistic with testing the significance of x in the regression of y on x and z.

x = matrix(rnorm(100, 0, 1),100,1)
colnames(x) <- paste0("X",1:ncol(x))
z = matrix(rbinom(100, 1, 0.5),100,1)
y = matrix(z[,1] + 2*x[,1] + rnorm(100, 0, 1), 100)
colnames(y) <- "y"
delta <- NULL

output <- parcorr.pvalue(factor.z=TRUE,x,y,z,delta,regression.type="linear")