LinearNormalizer

Normalizes the input matrix column-wise. The normalization can be executed so that

each column has the same mean, or
each column has the same median, or
each column has is fitted to some range specified by parameters, or
columns are normalized using z normalization, or
columns are z normalized using median instead of mean, or
columns are normalized to have unit variance.

The common mean level of a column can be configured with a parameter. Normalization is done by subtracting the actual mean of each column and adding the common mean level.

If a column contains missing values, the mean is computed from non-missing values.

Version	1.0
Bundle	tools
Categories	Normalization
Specialties	generic
Authors	Kari Nousiainen (Kari.Nousiainen@Helsinki.FI), Ville Rantanen (ville.rantanen@helsinki.fi)
Issue tracker	View/Report issues
Requires	R
Source files	component.xml LinearNormalizer.r
Usage	Example with default values LinearNormalizer( in = T, excludeColumns = "", group = "", lengthOfRange = 2, meanOfRange = 0, method = "mean" )

Type parameters (generics)

T (extends Matrix)

Inputs

Name	Type	Mandatory	Description
in	T (generic)	Mandatory	Input matrix.

Outputs

Name	Type	Description
out	T (generic)	Output matrix where each column has the same mean.

Parameters

Name	Type	Default	Description
excludeColumns	string	""	List of columns excluded from normalization. Normalizes all the numeric columns, so use this if your ID is numeric.
group	string	""	Label of the column that provides grouping. If given, the normalization is done individually for each unique group instead of the whole table.
lengthOfRange	float	2	Length of the range. This parameter is used only with range normalization.
meanOfRange	float	0	Common mean level. Each column will have a mean defined by this parameter when using method "mean". When using method "median" this value is added to every value after substraction by median. When using method "range" this parameters shows the mean of the range. This parameter is added to every value after normalization when using "z", "zMedian", or "unit variance" methods.
method	string	"mean"	Possible values are "mean", "median", "range", "z", "zMedian" and "unitVariance" normalizations.

Test cases

Test case	Parameters▼	IN in	OUT out
case1	(missing)	in	out
case2_median	properties	in	out
	method=median, meanOfRange=10, excludeColumns=Status,Expected
case3_grouping	properties	in	out
	method=mean, meanOfRange=0, group=Group

Generated 2019-02-08 07:42:18 by Anduril 2.0.0