genfis

Generate fuzzy inference system object from data

Syntax

fis = genfis(inputData,outputData)

fis = genfis(inputData,outputData,options)

Description

fis = genfis(inputData,outputData) returns a single-output Sugeno fuzzy inference system (FIS) using a grid partition of the given input and output data.

fis = genfis(inputData,outputData,options) returns an FIS generated using the specified input/output data and options. You can generate fuzzy systems using grid partitioning, subtractive clustering, or fuzzy c-means (FCM) clustering.

Examples

Generate Fuzzy Inference System Using Default Options

Define training data.

inputData = [rand(10,1) 10*rand(10,1)-5];
outputData = rand(10,1);

Generate a fuzzy inference system.

fis = genfis(inputData,outputData);

The generated system, fis, is created using grid partitioning with default options.

Generate FIS Using Grid Partitioning

Define training data.

inputData = [rand(10,1) 10*rand(10,1)-5];
outputData = rand(10,1);

Create a default genfisOptions option set for grid partitioning.

opt = genfisOptions('GridPartition');

Specify the following input membership functions for the generated FIS:

• 3 Gaussian membership functions for the first input variable

• 5 triangular membership functions for the second input variable

opt.NumMembershipFunctions = [3 5];
opt.InputMembershipFunctionType = ["gaussmf" "trimf"];

Generate the FIS.

fis = genfis(inputData,outputData,opt);

Plot the input membership functions. Each input variable has the specified number and type of input membership functions, evenly distributed over their input range.

[x,mf] = plotmf(fis,'input',1);
subplot(2,1,1)
plot(x,mf)
xlabel('input 1 (gaussmf)')
[x,mf] = plotmf(fis,'input',2);
subplot(2,1,2)
plot(x,mf)
xlabel('input 2 (trimf)')

Generate FIS Using Subtractive Clustering

Obtain input and output training data.

load clusterdemo.dat
inputData = clusterdemo(:,1:2);
outputData = clusterdemo(:,3);

Create a genfisOptions option set and specify the range of influence for each data dimension. Specify 0.5 and 0.25 as the range of influence for the first and second input variables. Specify 0.3 as the range of influence for the output data.

opt = genfisOptions('SubtractiveClustering',...
                    'ClusterInfluenceRange',[0.5 0.25 0.3]);

Generate the FIS.

fis = genfis(inputData,outputData,opt);

The generated FIS contains one rule for each cluster.

showrule(fis)

ans = 3x83 char array
    '1. If (in1 is in1cluster1) and (in2 is in2cluster1) then (out1 is out1cluster1) (1)'
    '2. If (in1 is in1cluster2) and (in2 is in2cluster2) then (out1 is out1cluster2) (1)'
    '3. If (in1 is in1cluster3) and (in2 is in2cluster3) then (out1 is out1cluster3) (1)'

Generate FIS Using FCM Clustering

Obtain the input and output data.

load clusterdemo.dat
inputData = clusterdemo(:,1:2);
outputData = clusterdemo(:,3);

Create a genfisOptions option set for FCM Clustering, specifying a Mamdani FIS type.

opt = genfisOptions('FCMClustering','FISType','mamdani');

Specify the number of clusters.

opt.NumClusters = 3;

Suppress the display of iteration information to the Command Window.

opt.Verbose = 0;

Generate the FIS.

fis = genfis(inputData,outputData,opt);

The generated FIS contains one rule for each cluster.

showrule(fis)

ans = 3x83 char array
    '1. If (in1 is in1cluster1) and (in2 is in2cluster1) then (out1 is out1cluster1) (1)'
    '2. If (in1 is in1cluster2) and (in2 is in2cluster2) then (out1 is out1cluster2) (1)'
    '3. If (in1 is in1cluster3) and (in2 is in2cluster3) then (out1 is out1cluster3) (1)'

Plot the input and output membership functions.

[x,mf] = plotmf(fis,'input',1);
subplot(3,1,1)
plot(x,mf)
xlabel('Membership Functions for Input 1')
[x,mf] = plotmf(fis,'input',2);
subplot(3,1,2)
plot(x,mf)
xlabel('Membership Functions for Input 2')
[x,mf] = plotmf(fis,'output',1);
subplot(3,1,3)
plot(x,mf)
xlabel('Membership Functions for Output')

Input Arguments

inputData — Input data

array

Input data, specified as an N-column array, where N is the number of FIS inputs.

inputData and outputData must have the same number of rows.

outputData — Output data

array

Output data, specified as an M-column array, where M is the number of FIS outputs.

When using grid partitioning, outputData must have one column. If you specify more than one column for grid partitioning, genfis uses the first column as the output data.

inputData and outputData must have the same number of rows.

options — FIS generation options

genfisOptions option set

FIS generation options, specified as a genfisOptions option set. If you do not specify options, genfis uses a default grid partitioning option set.

You can generate fuzzy systems using one of the following methods, which you specify when you create the option set:

• Grid partitioning — Generate input membership functions by uniformly partitioning the input variable ranges, and create a single-output Sugeno fuzzy system. The fuzzy rule base contains one rule for each input membership function combination.

  options = genfisOptions('GridPartition');

• Subtractive clustering — Generate a Sugeno fuzzy system using membership functions and rules derived from data clusters found using subtractive clustering of input and output data. For more information on subtractive clustering, see subclust.

  options = genfisOptions('SubtractiveClustering');

• FCM Clustering — Generate a fuzzy system using membership function and rules derived from data clusters found using FCM clustering of input and output data. For more information on FCM clustering, see fcm.

  options = genfisOptions('FCMClustering');

Output Arguments

fis — Fuzzy inference system

mamfis object | sugfis object

Fuzzy inference system, returned as a mamfis or sugfis object. The properties of fis depend on the type of clustering used and the corresponding options.

Clustering Type	Fuzzy System Type	Input Membership Functions	Fuzzy Rules	Output Membership Functions
Grid Partitioning	Sugeno	Each input variable has evenly distributed input membership function. Specify the number of membership functions using options.NumMembershipFunctions. Specify the membership function type using options.InputMembershipFunctionType.	One rule for each input membership function combination. The consequent of each rule corresponds to a different output membership function.	One output membership function for each fuzzy rule. Specify the membership function type using options.OutputMembershipFunctionType.
Subtractive Clustering	Sugeno	Each input variable has one 'gaussmf' input membership function for each fuzzy cluster.	One rule for each fuzzy cluster	Each output variable has one 'linear' output membership function for each fuzzy cluster.
FCM Clustering	Mamdani or Sugeno	Each input variable has one 'gaussmf' input membership function for each fuzzy cluster.	One rule for each fuzzy cluster	Each output variable has one output membership function for each fuzzy cluster. The membership function type is 'gaussmf' for Mamdani systems and 'linear' for Sugeno systems.

If fis is a single-output Sugeno system, you can tune the membership function parameters using the anfis command.

Compatibility Considerations

Support for representing fuzzy inference systems as structures will be removed

Not recommended starting in R2018b

Support for representing fuzzy inference systems as structures will be removed in a future release. Use mamfis and sugfis objects instead. There are differences between these representations that require updates to your code. These differences include:

• Object property names that differ from the corresponding structure fields

• Objects that store text data as strings rather than as character vectors

Also, all Fuzzy Logic Toolbox™ functions that accepted or returned fuzzy inference systems as structures now accept and return either mamfis or sugfis objects.

To convert existing fuzzy inference system structures to objects, use the convertfis function.