Numero (version 1.7.0)

nroStatistic: Estimate average value

Description

Calculate weighted descriptive statistics for datasets with missing and duplicated values.

Usage

nroStatistic(data, weights = NULL, method = "center")

Arguments

data

A matrix or a data frame.

weights

A matrix or a data frame of positive values with the same row and column names as data.

method

The name of the test statistic, see details.

Value

A numeric vector with elements named according to data columns.

Details

Methods include "center", "mean", "median", "mode", "min", "max", "range", "sd","iqr", "var" and "number".

For binary columns, the method "center" returns the mean. If there are more than two distinct values (nuniq > 2), the formula rho * mean + (1 - rho) * median, rho = 2.2 / nuniq / log(nuniq + 1) is used. If a column contains duplicated values, the duplicates are expanded linearly prior to calculating the median in the formula. This mitigates numerical artefacts from bootstrapping or permutations.

Range is defined the difference between the maximum and minimum value and "iqr" stands for inter-quartile range.

The method "number" returns the number of usable values.

Examples

# NOT RUN {
# Import data.
fname <- system.file("extdata", "finndiane.txt", package = "Numero")
dataset <- read.delim(file = fname)

# Calculate centers.
mu <- nroStatistic(dataset, method = "center")
print(mu)

# Calculate inter-quartile ranges.
iqr <- nroStatistic(dataset, method = "iqr")
print(iqr)
# }