RcppColMetric: Efficient Column-Wise Metric Computation Against Common Vector

Description

In data science, it is a common practice to compute a series of columns (e.g. features) against a common response vector. Various metrics are provided with efficient computation implemented with 'Rcpp'.

Author(s)

Maintainer: Xiurui Zhu zxr6@163.com

See Also

Useful links:

• https://github.com/zhuxr11/RcppColMetric

• Report bugs at https://github.com/zhuxr11/RcppColMetric/issues

Column-wise area under ROC curve (AUC)

Description

Calculate area under the ROC curve (AUC) for every column of a matrix or data frame. For better performance, data frame is preferred.

Usage

col_auc(x, y, args = NULL)

Arguments

Value

An output is a single matrix with the same number of columns as X and "n choose 2" ( n!/((n-2)! 2!) = n(n-1)/2 ) number of rows, where n is number of unique labels in y list. For example, if y contains only two unique class labels ( length(unique(lab))==2 ) then output matrix will have a single row containing AUC of each column. If more than two unique labels are present than AUC is calculated for every possible pairing of classes ("n choose 2" of them).

Note

Change log:

• 0.1.0 Xiurui Zhu - Initiate the function.

See Also

caTools::colAUC for the original R implementation.

col_auc_vec for the vectorized version.

Examples

if (require("MASS", quietly = TRUE) == TRUE) {
  data(cats)
  print(res_cpp <- col_auc(cats[, 2L:3L], cats[, 1L]))
  # Validate with caTools::colAUC()
  if (require("caTools", quietly = TRUE) == TRUE) {
    print(res_r <- caTools::colAUC(cats[, 2L:3L], cats[, 1L]))
    identical(res_cpp, res_r)
  }
}

Vectorized version of function col_auc

Description

This is the vectorized version of col_auc.

Usage

col_auc_vec(x, y, args = NULL)

Arguments

Value

List, where each element is an output from col_auc.

Note

Change log:

• 0.1.0 Xiurui Zhu - Initiate the function.

See Also

col_auc for the non-vectorized version.

Examples

if (require("MASS", quietly = TRUE) == TRUE) {
  data(cats)
  print(res_cpp <- col_auc_vec(list(cats[, 2L:3L]), list(cats[, 1L])))
  # Validate with caTools::colAUC()
  if (require("caTools", quietly = TRUE) == TRUE) {
    print(res_r <- caTools::colAUC(cats[, 2L:3L], cats[, 1L]))
    identical(res_cpp, list(res_r))
  }
}

Column-wise mutual information

Description

Calculate mutual information for every column of a matrix or data frame. Only discrete values are allowed. For better performance, data frame is preferred.

Usage

col_mut_info(x, y, args = NULL)

Arguments

Value

An output is a single matrix with the same number of columns as X and 1 row.

Note

Change log:

• 0.1.0 Xiurui Zhu - Initiate the function.

See Also

infotheo::mutinformation for the original computation of mutual information in R (and also the computation methods).

col_mut_info_vec for the vectorized version.

Examples

if (require("MASS", quietly = TRUE) == TRUE) {
  data(cats)
  print(res_cpp <- col_mut_info(round(cats[, 2L:3L]), cats[, 1L]))
  # Validate with caTools::colAUC()
  if ((require("infotheo", quietly = TRUE) == TRUE) &&
      (require("magrittr", quietly = TRUE) == TRUE)) {
    print(res_r <- sapply(round(cats[, 2L:3L]), infotheo::mutinformation, cats[, 1L]) %>%
            {matrix(., nrow = 1L, dimnames = list(NULL, names(.)))})
    identical(res_cpp, res_r)
  }
}

Vectorized version of function col_mut_info

Description

This is the vectorized version of col_mut_info.

Usage

col_mut_info_vec(x, y, args = NULL)

Arguments

Value

List, where each element is an output from col_mut_info.

Note

Change log:

• 0.1.0 Xiurui Zhu - Initiate the function.

See Also

col_mut_info for the non-vectorized version.

Examples

if (require("MASS", quietly = TRUE) == TRUE) {
  data(cats)
  print(res_cpp <- col_mut_info_vec(list(round(cats[, 2L:3L])), list(cats[, 1L])))
  # Validate with caTools::colAUC()
  if ((require("infotheo", quietly = TRUE) == TRUE) &&
      (require("magrittr", quietly = TRUE) == TRUE)) {
    print(res_r <- sapply(round(cats[, 2L:3L]), infotheo::mutinformation, cats[, 1L]) %>%
            {matrix(., nrow = 1L, dimnames = list(NULL, names(.)))})
    identical(res_cpp, list(res_r))
  }
}