Chi square matrix residuals-CodePudding

below is a function to extract p-values from multiple Chi-Square tests and display them as a matrix. I'm trying to do the same, but to extract residuals instead. Any help is appreciated.

Sample data:

df <- data.frame(first_column  = c(rep("E1_C1",5), rep("E1_C2",3), rep("E2_C2",7),rep("E3_C3",5)),
                  second_column = c(rep("E1_C1",3), rep("E1_C2",10), rep("E2_C2",4),rep("E3_C3",3)),
                  third_column = c(rep("E1_C1",7), rep("E1_C2",4), rep("E2_C2",3),rep("E3_C3",6)),
                  fourth_column = c(rep("E1_C1",4), rep("E1_C2",6), rep("E2_C2",6),rep("E3_C3",4))
)

Chi-square matrix function for P-Values:

chisqmatrix <- function(x) {
  names = colnames(x);  num = length(names)
  m = matrix(nrow=num,ncol=num,dimnames=list(names,names))
  for (i in 1:(num-1)) {
    for (j in (i 1):num) {
      #browser()
      m[j,i] = chisq.test(x[, i, drop = TRUE],x[, j, drop = TRUE])$p.value
    }
  }
  return (m)
}

CodePudding user response：

In your case, the returned residuals is a 4x4 matrix. Instead of using a matrix to take the results, the following solution uses a list instead. This way you can have matrices of different sizes.

With minimal changes from your original code:

chisqlist <- function(x) {
  names = colnames(x);  num = length(names)
  m = list()
  index = 1
  for (i in 1:(num-1)) {
    for (j in (i 1):num) {
      #browser()
      m[[index]] = chisq.test(x[, i, drop = TRUE],x[, j, drop = TRUE])$residuals
      index=index 1
    }
  }
  return (m)
}

Edit: I do prefer @ Onyambu's answer, which I didn't see. It would be faster than a nested for loop.

CodePudding user response：

Simply change your function from requesting $p.value to requesting $residuals. This will provide (observed - expected) / sqrt(expected). If you desire standardized residuals request $stdres.

  chisqmatrix <- function(x) {
    names = colnames(x);  num = length(names)
    m = matrix(nrow=num,ncol=num,dimnames=list(names,names))
    for (i in 1:(num-1)) {
      for (j in (i 1):num) {
        #browser()
        m[j,i] = chisq.test(x[, i, drop = TRUE],x[, j, drop = TRUE])$residuals
      }
    }
    return (m)
  }