optimizableVariable

Variable description for bayesopt or other optimizers

Description

Create variables for optimizers.

Creation

Syntax

variable = optimizableVariable(Name,Range)

variable = optimizableVariable(Name,Range,Name,Value)

Description

variable = optimizableVariable(Name,Range) creates a variable with the specified name and range of values.

variable = optimizableVariable(Name,Range,Name,Value) sets properties using name-value arguments. For example, optimizableVariable('xvar',[1 1000],'Type','integer') creates an integer variable from 1 to 1000. You can specify multiple name-value arguments. Enclose each property name in quotes.

Properties

Name — Variable name

character vector | string scalar

Variable name, specified as a character vector or string scalar. The name must be unique, meaning different from those of other variables in the optimization.

Note

• There are two names associated with an optimizableVariable:

  • The MATLAB^® workspace variable name

  • The name of the variable in the optimization

  For example,

  xvar = optimizableVariable('spacevar',[1,100]);

  xvar is the MATLAB workspace variable, and 'spacevar' is the variable in the optimization.

  Use these names as follows:

  • Use xvar as an element in the vector of variables you pass to bayesopt. For example,

    results = bayesopt(fun,[xvar,tvar])

  • Use 'spacevar' as the name of the variable in the optimization. For example, in an objective function,

    function objective = mysvmfun(x,cdata,grp)
    SVMModel = fitcsvm(cdata,grp,'KernelFunction','rbf',...
        'BoxConstraint',x.spacevar,...
        'KernelScale',x.tvar);
    objective = kfoldLoss(crossval(SVMModel));

Example: 'X1'

Data Types: char | string

Range — Variable range

2-element increasing real vector | string array or cell array of category names

Variable range, specified as a 2-element finite increasing real vector, or as a string array or cell array of category names:

• For real or integer variables, Range gives the lower bound and upper bound of that variable.

• For categorical variables, Range gives the possible values.

Example: [-10,1]

Example: {'red','blue','black'}

Data Types: double | string | cell

Type — Variable type

'real' (default) | 'integer' | 'categorical'

Variable type, specified as 'real' (real variable), 'integer' (integer variable), or 'categorical' (categorical variable).

Note

The MATLAB data type of both 'real' and 'integer' variables is the standard double-precision floating point number. The data type of 'categorical' variables is categorical. So, for example, to read a value of a categorical variable named 'colorv' in a table of variables named x, use the command char(x.colorv). For an example, see the objective function in Custom Output Functions.

Example: 'Type','categorical'

Transform — Transform applied to variable

'none' (default) | 'log'

Transform applied to the variable, specified as 'none' (no transform) or 'log' (logarithmic transform).

For 'log', the variable must be a positive real variable ('Type','real') or a nonnegative integer variable ('Type','integer'). The software searches and models the variable on a log scale.

Example: 'Transform','log'

Optimize — Indication to use variable in optimization

true (default) | false

Indication to use variable in optimization, specified as true (use the variable) or false (do not use the variable).

Example: 'Optimize',false

Data Types: logical

Note

You can use dot notation to change the following properties after creation.

• Range of real or integer variables. For example,

  xvar = optimizableVariable('x',[-10,10]);
  % Modify the range:
  xvar.Range = [1,5];

• Type between 'integer' and 'real'. For example,

  xvar.Type = 'integer';

• Transform of real or integer variables between 'log' and 'none'. For example,

  xvar.Transform = 'log';

You can use this flexibility, for example, to tweak an optimization that you want to continue. Update the range or transform using dot notation and then call resume.

Object Functions

bayesopt

Select optimal machine learning hyperparameters using Bayesian optimization

Examples

Variables for Optimization Examples

Real variable from 0 to 1:

var1 = optimizableVariable('xvar',[0 1])

var1 = 
  optimizableVariable with properties:

         Name: 'xvar'
        Range: [0 1]
         Type: 'real'
    Transform: 'none'
     Optimize: 1

Integer variable from 0 to 1000 on a log scale:

var2 = optimizableVariable('ivar',[0 1000],'Type','integer','Transform','log')

var2 = 
  optimizableVariable with properties:

         Name: 'ivar'
        Range: [0 1000]
         Type: 'integer'
    Transform: 'log'
     Optimize: 1

Categorical variable of rainbow colors:

var3 = optimizableVariable('rvar',{'r' 'o' 'y' 'g' 'b' 'i' 'v'},'Type','categorical')

var3 = 
  optimizableVariable with properties:

         Name: 'rvar'
        Range: {'r'  'o'  'y'  'g'  'b'  'i'  'v'}
         Type: 'categorical'
    Transform: 'none'
     Optimize: 1

Version History

Introduced in R2016b