boot_simulated_cat_bin {foodingraph}

Confidence-interval bootstraps on simulated independent variables

Description

Create a defined number of simulated independent random variables of a given size according to type : 2 ordinal variables, 2 binary variables, 1 binary and 1 ordinal variable. A number of bootstraps are then performed on the sample to calculate a confidence interval of the bootstrap distribution of the chosen method: mutual information or the maximal information coefficient. The percentile method is used to calculate this interval.

Usage

boot_simulated_cat_bin(type = c("cat", "bin", "bincat"),
  method = c("mic", "mi"), simu = 10, boots = 5000, size = 500,
  percentile = 0.99)

Arguments

type

: the type of the simulated variables: cat is for 2 ordinal variables, bin is for 2 binary variables, bincat is for 1 binary and 1 ordinal variable.

method

: the method used to calculate the association : mutual information (mi), or the maximal information coefficient (mic).

simu

: the number of simulated pairs of variables. For each pair, the confidence-interval bootstrap is calculated from the bootstrap distribution of the MI/MIC of between the two pairs. At the end of the program, the mean of the chosen percentile is given. Default is 10.

boots

: the number of bootstraps per simulation. Default is 5000.

size

: the size of the sample. Default is 500.

percentile

: the percentile kept. Default is 0.99 (the 99th percentile).

Value

The mean of the percentile values.

References

Reshef et al. (2011) <doi:10.1126/science.1205438>

Meyer et al. (2008) <doi:10.1186/1471-2105-9-461>

Examples

boot_simulated_cat_bin("cat", "mic", 2, 500)