Reorder Category Levels

Note

The nominal and ordinal array data types are not recommended. To represent ordered and unordered discrete, nonnumeric data, use the Categorical Arrays data type instead.

• Reorder Category Levels in Ordinal Arrays

• Reorder Category Levels in Nominal Arrays

Reorder Category Levels in Ordinal Arrays

This example shows how to reorder the category levels in an ordinal array using reorderlevels.

Load sample data.

AllSizes = {'medium','large','small','small','medium',...
            'large','medium','small'};

The created variable, AllSizes, is a cell array of character vectors containing size measurements on eight objects.

Create an ordinal array.

Convert AllSizes to an ordinal array without specifying the order of the category levels.

size = ordinal(AllSizes);
getlevels(size)

ans = 1x3 ordinal
     large      medium      small 

By default, the categories are ordered by their labels in ascending alphabetical order, large < medium < small.

Compare elements.

Check whether or not the first object (which has size medium) is smaller than the second object (which has size large).

size(1) < size(2)

ans = logical
   0

The logical value 0 indicates that the medium object is not smaller than the large object.

Reorder category levels.

Reorder the category levels so that small < medium < large.

size = reorderlevels(size,{'small','medium','large'});
getlevels(size)

ans = 1x3 ordinal
     small      medium      large 

Compare elements.

Verify that the first object is now smaller than the second object.

size(1) < size(2)

ans = logical
   1

The logical value 1 indicates that the expected inequality now holds.

Reorder Category Levels in Nominal Arrays

This example shows how to reorder the category levels in nominal arrays using reorderlevels. By definition, nominal array categories have no natural ordering. However, you might want to change the order of levels for display or analysis purposes. For example, when fitting a regression model with categorical covariates, fitlm uses the first level of a nominal independent variable as the reference category.

Load sample data.

The dataset array, hospital, contains variables measured on 100 sample patients. The variable Weight contains the weight of each patient. The variable Sex is a nominal variable containing the gender, Male or Female, for each patient.

load hospital
getlevels(hospital.Sex)

ans = 1x2 nominal
     Female      Male 

By default, the order of the nominal categories is in ascending alphabetical order of the labels.

Plot data grouped by category level.

Draw box plots of weight, grouped by gender.

figure
boxplot(hospital.Weight,hospital.Sex)
title('Weight by Gender')

The box plots appear in the same alphabetical order returned by getlevels.

Change the category order.

Change the order of the category levels.

hospital.Sex = reorderlevels(hospital.Sex,{'Male','Female'});
getlevels(hospital.Sex)

ans = 1x2 nominal
     Male      Female 

The levels are in the newly specified order.

Plot data in new order.

Draw box plots of weight by gender.

figure
boxplot(hospital.Weight,hospital.Sex)
title('Weight by Gender')

The order of the box plots corresponds to the new level order.