Typo in interestMeasure when measure is chiSquared

[bachnguyen-tomo]

The documentation mentioned that using chiSquared as an interest measure supports mining substitution effects, by setting the parametercomplement = TRUE, however it's actually "compliment" in interestMeasure.R leading to this error message
Error in .basicRuleMeasure(x, measure, transactions = transactions, ...) (quarterly.R#23): unused argument (complement = TRUE)

 chiSquared = {
        chi2 <- numeric(length(x))
        
        for (i in seq_len(length(x))) {
          fo <- matrix(c(n00[i], n01[i], n10[i], n11[i]), ncol = 2)
          #fe <- tcrossprod(c(nx0[i], nx1[i]), ic(n0x[i], n1x[i])) / n
          ## check if approximation is ok
          ## we don't do this now
          ##if(any(fe < 5)) chi2[i] <- nA
          ##else
          #chi2[i] <- sum((fo - fe) ^ 2 / fe)
          
          # warning about approximation
          suppressWarnings(chi2[i] <-
              stats::chisq.test(fo, correct = FALSE)$statistic)
        }
        
        ## the chi square test has 1 df for a 2x2 contingency table.
        ## The critical value at alpha=0.05 is:
        ## qchisq(0.05, df =1, lower.tail=FALSE)
        ## [1] 3.841459
        if (!significance)
          chi2
        else
          stats::pchisq(q = chi2,
            df = 1,
            lower.tail = !compliment)
      }
    ))

due to the variable declared by the bigger class as

.basicRuleMeasure <- function(x,
  measure,
  transactions = NULL,
  ### adds smoothCounts to the count in each cell to avoid counts of 0
  smoothCounts = 0,
  ### k is the number of classes used by laplace
  significance = FALSE,
  ### used by chi-squared
  compliment = TRUE,
...)

fishersExactTest and hyperConfidence supports mining substitution effect, but they have their own "complement" parameter