Calculate Lin's concordance correlation coefficient

Calculates the concordance correlation coefficient as derived by Lin for a set of matched observations in multiple classes.

Usage

calculate.concordance.coefficient(data.mat);

Arguments

data.mat

A matrix of numeric data, with columns representing classes and row representing matched observations

Details

This function calculates the Lin concordance correlation coefficient for a set of matched observations for any number of observers. Roughly speaking, this metric measures the orthogonal distance of the data from the y=x line. For further information on Lin's methods, see 'A concordance correlation coefficient to evaluate reproducability', Biometrics 1989, 45:1, pp. 255-68 (Section 7) and 'A note on the concordance correlation coefficient', Biometrics 2000, 56:1, pp. 324-325 (A correction to the formula given in the former).

Value

Returns a single value on the interval [-1,1] representing the degree to which the classes are in concordance with one another (interpreted the same as a Pearson or a Spearman correlation coefficient).

Author

Michal Grzadkowski

Examples

# a matrix with four classes of 100 random observations each
data.matrix1 <- matrix(
      runif(400),
      ncol = 4
      );
calculate.concordance.coefficient(data.matrix1);
#> [1] 0.01048962

# a matrix with four classes of 100 similar observations each
base.data <- sample(
    x = 0:1000,
    size = 100,
    replace = TRUE
    );
data.matrix2 <- matrix(
      c(
        base.data,
        base.data + runif(100),
        base.data + runif(100),
        base.data + runif(100)
        ),
      ncol = 4
      );
calculate.concordance.coefficient(data.matrix2);
#> [1] 0.9999989