Movatterモバイル変換

FisPro package

Description

This package is a basic implementation of the main functions to use a "Fuzzy Inference System" that can be used for reasoning purposes, especially for simulating a physical or biological system. It is derived from theFisPro open source software. Fuzzy inference systems are briefly described in theFuzzy Logic Elementary Glossary. They are based on fuzzy rules, which have a good capability for managing progressive phenomenons. Fuzzy logic, since the pioneer work by Zadeh, has proven to be a powerful interface between symbolic and numerical spaces. One of the reasons for this success is the ability of fuzzy systems to incorporate human expert knowledge with its nuances, as well as to express the behaviour of the system in an interpretable way for humans. Another reason is the possibility of designing data-driven FIS to make the most of available data.
To design a fuzzy system that can be handled by this package the user can use theFisPro software. If needed, the package can be extended to other functions.
All the mentioned publications are available from theFisPro web site.
EnjoyFisPro!

Author(s)

FisPro Teamcontact@fispro.org

References

Guillaume S, Charnomordic B (2011).“Learning interpretable Fuzzy Inference Systems with FisPro.”International Journal of Information Sciences,181(20), 4409-4427.doi:10.1016/j.ins.2011.03.025, Special Issue on Interpretable Fuzzy Systems.

Guillaume S, Charnomordic B (2012).“Fuzzy Inference Systems: an integrated modelling environment for collaboration between expert knowledge and data using FisPro.”Expert Systems with Applications,39(10), 8744-8755.doi:10.1016/j.eswa.2012.01.206.

See Also

https://www.fispro.org

Class "Fis"

Description

Class to manage a Fis "Fuzzy Inference System"

Fields

name

character vector, The name of the Fis

conjunction

character vector, The conjunction operator of rules in the Fis
Allowed values are: "min" (the default), "prod" or "luka"

Constructors

Fis()

The default constructor to build an empty Fis
The Fis is initialized with "min" conjunction and empty name
The design must be completed using the available functions to add inputs, outputs and rules before it can be used for inference

return:

Fis object

Fis(fis_file)

The constructor to build a Fis from a configuration file
The configuration file can be designed using theFisPro open source software

argument:fis_file

character vector, The filename of the Fis configuration file

return:

Fis object

Methods

input_size()

return:

integer value, The number of inputs in the Fis

add_input(input)

argument:input

FisIn object, The input to add in the Fis

get_input(input_index)

argument:input_index

integer value, The index (1-based index) of the input in the Fis

return:

FisIn object

get_inputs()

Get all inputs in the Fis

return:

list ofFisIn objects

output_size()

return:

integer value, The number of outputs in the Fis

add_output(output)

argument:output

FisOut object, The output to add in the Fis

get_output(output_index)

argument:output_index

integer value, The index (1-based index) of the output in the Fis

return:

FisOut object

get_outputs()

Get all outputs in the Fis

return:

list ofFisOut objects

rule_size()

return:

integer value, The number of rules in the Fis

add_rule(rule)

argument:rule

Rule object, The rule to add in the Fis

get_rule(rule_index)

argument:rule_index

integer value, The index (1-based index) of the rule in the Fis

return:

Rule object

get_rules()

Get all rules in the Fis

return:

list ofRule objects

infer(data)

Infers all outputs

argument:data

numeric vector,matrix ordata.frame, The input data or dataset to infer (the vector length or the number of columns must be equal to the number of inputs)

return:

numeric vector ormatrix (in case of 2D input data)

infer_output(data, output_index)

Infers a single output

argument:data

numeric vector,matrix ordata.frame, The input data or dataset to infer (the vector length or the number of columns must be equal to the number of inputs)

argument:output_index

integer value, The index (1-based index) of the output to infer

return:

numeric value or vector (in case of 2D input data)

See Also

NewFis

Fuzzy Logic Elementary Glossary

Examples

# build a Fis from a configuration filefis_file <- system.file("extdata", "test.fis", package = "FisPro")fis <- NewFis(fis_file)# infers all outputsinferred <- fis$infer(c(0.25, 0.75))# infers first outputinferred_output1 <- fis$infer_output(c(0.25, 0.75), 1)# infers second outputinferred_output2 <- fis$infer_output(c(0.25, 0.75), 2)# infers test_data datasettest_file <- system.file("extdata", "test_data.csv", package = "FisPro")dataset <- read.csv(test_file)inferred_dataset <- fis$infer(dataset)######################################################################### or build a Fis from scratchfis <- NewFis()fis$name <- "foo"# build the first inputfisin1 <- NewFisIn(0, 1)fisin1$name <- "input1"fisin1$add_mf(NewMfTrapezoidalInf(0, 1))fisin1$add_mf(NewMfTrapezoidalSup(0, 1))fis$add_input(fisin1)# build the second inputfisin2 <- NewFisIn(0, 1)fisin2$name <- "input2"fisin2$add_mf(NewMfTrapezoidalInf(0, 0.5))fisin2$add_mf(NewMfTriangular(0, 0.5, 1))fisin2$add_mf(NewMfTrapezoidalSup(0.5, 1))fis$add_input(fisin2)# build an outputfisout <- NewFisOutCrisp(0, 1)fisout$name <- "output"fis$add_output(fisout)# add rules to the Fisfis$add_rule(NewRule(c(1, 2), 0))fis$add_rule(NewRule(c(2, 0), 1))

Class "Fisin"

Description

Class to manage aFis input

Fields

name

character vector, The name of the input

Constructors

FisIn()

The default constructor to build an empty input with the default range [0, 1]

return:

FisIn object

FisIn(minimum, maximum)

The constructor to build an empty input

argument:minimum

numeric value, The minimum range value of the input

argument:maximum

numeric value, The maximum range value of the input

return:

FisIn object

FisIn(number_of_mfs, minimum, maximum)

The constructor to build an input with a regular standardized fuzzy partition

argument:number_of_mfs

integer value, The number of Mfs in the fuzzy partition

argument:minimum

numeric value, The minimum range value of the input

argument:maximum

numeric value, The maximum range value of the input

return:

FisIn object

FisIn(breakpoints, minimum, maximum)

The constructor to build an input with an irregular standardized fuzzy partition

argument:breakpoints

numeric vector, The breakpoint values (sorted in ascending order) of the Mfs in the fuzzy partition

argument:minimum

numeric value, The minimum range value of the input

argument:maximum

numeric value, The maximum range value of the input

return:

FisIn object

Methods

range()

return:

numeric vector, The range of the input (min max values)

mf_size()

return:

integer value, The number of Mfs in the input partition

add_mf(mf)

Add an Mf in the input partition

argument:mf

Mf object, The Mf to add

get_mf(mf_index)

argument:mf_index

integer value, The index (1-based index) of the mf to return

return:

Mf object

get_mfs()

Get all mfs in the input

return:

list ofMf objects

is_standardized()

return:

logical value,TRUE if the input is a standardized fuzzy partition,FALSE otherwise

See Also

NewFisIn

Fuzzy Logic Elementary Glossary

Examples

input <- NewFisIn(0, 2)input$name <- "foo"input$add_mf(NewMfTrapezoidalInf(0, 1))input$add_mf(NewMfTriangular(0, 1, 2))input$add_mf(NewMfTrapezoidalSup(1, 2))

Class "FisOut"

Description

The base class ofFis output (cannot be instantiate)
Use derived classesFisOutCrisp orFisOutFuzzy

Fields

name

character vector, The name of the output

Methods

range()

return:

numeric vector, The range of the output (min max values)

Class "FisOutCrisp"

Description

Class to manage aFis crisp output

Fields

defuzzification

character vector, The defuzzification operator of the crisp output
Allowed values are: "sugeno" (the default) or "MaxCrisp"

disjunction

character vector, The disjunction operator of the crisp output
Allowed values are: "max" (the default) or "sum"

Inherits

FisOutCrisp class inherits all fields and methods ofFisOut class

Constructors

FisOutCrisp()

The default constructor to build a crisp output with the default range [0, 1]

return:

FisOutCrisp object

FisOutCrisp(minimum, maximum)

The constructor to build a crisp output

argument:minimum

numeric value, The minimum range value of the output

argument:maximum

numeric value, The maximum range value of the output

return:

FisOutCrisp object

See Also

NewFisOutCrisp

Fuzzy Logic Elementary Glossary

Examples

output <- NewFisOutCrisp(0, 1)output$name <- "foo"output$defuzzification <- "sugeno"output$disjunction <- "max"

Class "FisOutFuzzy"

Description

Class to manage aFis fuzzy output

Fields

defuzzification

character vector, The defuzzification operator of the fuzzy output
Allowed values are: "sugeno" (the default) "MeanMax", or "area"

disjunction

character vector, The disjunction operator of the fuzzy output
Allowed values are: "max" (the default) or "sum"

Inherits

FisOutFuzzy class inherits all fields and methods ofFisOut class

Constructors

FisOutFuzzy()

The default constructor to build a fuzzy output with the default range [0, 1]

return:

FisOutFuzzy object

FisOutFuzzy(minimum, maximum)

The constructor to build a fuzzy output

argument:minimum

numeric value, The minimum range value of the output

argument:maximum

numeric value, The maximum range value of the output

return:

FisOutFuzzy object

FisOutFuzzy(number_of_mfs, minimum, maximum)

The constructor to build a fuzzy with a regular standardized fuzzy partition

argument:number_of_mfs

integer value, The number of Mfs in the fuzzy partition

argument:minimum

numeric value, The minimum range value of the output

argument:maximum

numeric value, The maximum range value of the output

return:

FisOutFuzzy object

FisOutFuzzy(breakpoints, minimum, maximum)

The constructor to build a fuzzy with an irregular standardized fuzzy partition

argument:breakpoints

numeric vector, The breakpoint values (sorted in ascending order) of the Mfs in the fuzzy partition

argument:minimum

numeric value, The minimum range value of the output

argument:maximum

numeric value, The maximum range value of the output

return:

FisOutFuzzy object

Methods

mf_size()

return:

integer value, The number of Mfs in the output partition

add_mf(mf)

Add an Mf in the output partition

argument:mf

Mf object, The Mf to add

get_mf(mf_index)

argument:mf_index

integer value, The index (1-based index) of the mf to return

return:

Mf object

get_mfs()

Get all mfs in the output

return:

list ofMf objects

is_standardized()

return:

logical value,TRUE if the output is a standardized fuzzy partition,FALSE otherwise

See Also

NewFisOutFuzzy

Fuzzy Logic Elementary Glossary

Examples

output <- NewFisOutFuzzy(0, 2)output$name <- "foo"output$defuzzification <- "sugeno"output$disjunction <- "max"output$add_mf(NewMfTrapezoidalInf(0, 1))output$add_mf(NewMfTriangular(0, 1, 2))output$add_mf(NewMfTrapezoidalSup(1, 2))

Deprecated classes in packageFisPro.

Description

The classes listed below are deprecated and will be removed in future version.Alternative classes with similar functionality are also mentioned.
Help pages for deprecated functions are available athelp("<class>-deprecated").

Deprecated classes

Forfis, useFis instead
Formf, useMf instead
Formf_triangular, useMfTriangular instead
Formf_trapezoidal_inf, useMfTrapezoidalInf instead
Formf_trapezoidal_sup, useMfTrapezoidalSup instead
Formf_trapezoidal, useMfTrapezoidal instead

See Also

fis-deprecated
mf-deprecated
mf_triangular-deprecated
mf_trapezoidal_inf-deprecated
mf_trapezoidal_sup-deprecated
mf_trapezoidal-deprecated

Class "Mf"

Description

The base class of all "membership function" classes (cannot be instantiate)
Use derived classesMfTriangular,MfTrapezoidal,MfTrapezoidalInf orMfTrapezoidalSup

Fields

label

character vector, The label of the membership function

Methods

degree(value)

Get the membership degree

argument:value

numeric value to compute the membership degree

return:

numeric value

See Also

Fuzzy Logic Elementary Glossary

Class "MfTrapezoidal"

Description

Class to manage a trapezoidal membership function

Inherits

MfTrapezoidal class inherits all fields and methods ofMf class

Constructors

MfTrapezoidal(lower_support, lower_kernel, upper_kernel, upper_support)

argument:lower_support

numeric lower value of support

argument:lower_kernel

numeric lower value of kernel

argument:upper_kernel

numeric upper value of kernel

argument:upper_support

numeric upper value of support

return:

MfTrapezoidal object

See Also

NewMfTrapezoidal

Examples

mf <- NewMfTrapezoidal(0, 1, 2, 3)mf$degree(0.5)

Class "MfTrapezoidalInf"

Description

Class to manage a trapezoidal inf membership function

Inherits

MfTrapezoidalInf class inherits all fields and methods ofMf class

Constructors

MfTrapezoidalInf(upper_kernel, upper_support)

argument:upper_kernel

numeric upper value of kernel

argument:upper_support

numeric upper value of support

return:

MfTrapezoidalInf object

See Also

NewMfTrapezoidalInf

Examples

mf <- NewMfTrapezoidalInf(0, 1)mf$degree(0.5)

Class "MfTrapezoidalSup"

Description

Class to manage a trapezoidal sup membership function

Inherits

MfTrapezoidalSup class inherits all fields and methods ofMf class

Constructors

MfTrapezoidalSup(lower_support, lower_kernel)

argument:lower_support

numeric lower value of support

argument:lower_kernel

numeric lower value of kernel

return:

MfTrapezoidalSup object

See Also

NewMfTrapezoidalSup

Examples

mf <- NewMfTrapezoidalSup(0, 1)mf$degree(0.5)

Class "MfTriangular"

Description

Class to manage a triangular membership function

Inherits

MfTriangular class inherits all fields and methods ofMf class

Constructors

MfTriangular(lower_support, kernel, upper_support)

argument:lower_support

numeric lower value of support

argument:kernel

numeric value of kernel

argument:upper_support

numeric upper value of support

return:

MfTriangular object

See Also

NewMfTriangular

Examples

mf <- NewMfTriangular(0, 1, 2)mf$degree(0.5)

Create object of class "Fis"

Description

Function to create object of classFis

Usage

NewFis(...)

Arguments

Value

Fis object

Create object of class "FisIn"

Description

Function to create object of classFisIn

Usage

NewFisIn(...)

Arguments

Value

FisIn object

Create object of class "FisOutCrisp"

Description

Function to create object of classFisOutCrisp

Usage

NewFisOutCrisp(...)

Arguments

Value

FisOutCrisp object

Create object of class "FisOutFuzzy"

Description

Function to create object of classFisOutFuzzy

Usage

NewFisOutFuzzy(...)

Arguments

Value

FisOutFuzzy object

Create object of class "MfTrapezoidal"

Description

Function to create object of classMfTrapezoidal

Usage

NewMfTrapezoidal(...)

Arguments

Value

MfTrapezoidal object

Create object of class "MfTrapezoidalInf"

Description

Function to create object of classMfTrapezoidalInf

Usage

NewMfTrapezoidalInf(...)

Arguments

Value

MfTrapezoidalInf object

Create object of class "MfTrapezoidalSup"

Description

Function to create object of classMfTrapezoidalSup

Usage

NewMfTrapezoidalSup(...)

Arguments

Value

MfTrapezoidalSup object

Create object of class "MfTriangular"

Description

Function to create object of classMfTriangular

Usage

NewMfTriangular(...)

Arguments

Value

MfTriangular object

Create object of class "Rule"

Description

Function to create object of classRule

Usage

NewRule(...)

Arguments

Value

Rule object

Class "Rule"

Description

Class to manage aFis rule

Fields

premises

integer vector, The premises of the rule
A premise is the 1-based index of MF in theFisIn
0 means the input is not taken into account for this rule, i.e. the rule is incomplete
The vector length must be equal to the number of inputs in theFis

conclusions

numeric vector, The conclusions of the rule
A conclusion is anumeric value for crisp outputFisOutCrisp, or the 1-based index of MF in the fuzzy outputFisOutFuzzy
The vector length must be equal to the number of outputs in theFis

Constructors

Rule()

The default constructor to build an empty rule
The rule is initialized with empty premises and conclusions

return:

Rule object

Rule(premises, conclusions)

The constructor to build a rule

argument:premises

integer vector, The premises of the rule (the vector length must be equal to the number of inputs in theFis)

argument:conclusions

numeric vector, The conclusions of the rule (the vector length must be equal to the number of outputs in theFis)

return:

Rule object

See Also

NewRule

Fuzzy Logic Elementary Glossary

Examples

rule1 <- NewRule()rule1$premises <- c(1, 2, 0)rule1$conclusions <- c(1, 2)rule2 <- NewRule(c(2, 1, 1), c(2, 1))

Deprecated class "fis"

Description

Thefis class is deprecated and will be removed in future version, useFis instead

Methods

infer_output(data, output_index)

WARNING !!! theoutput_index is now 1-based indexed inFis, was 0-based indexed infis

See Also

FisPro-deprecated

Deprecated class "mf"

Description

Themf class is deprecated and will be removed in future version, useMf instead

See Also

FisPro-deprecated

Deprecated class "mf_trapezoidal"

Description

Themf_trapezoidal class is deprecated and will be removed in future version, useMfTrapezoidal instead

See Also

FisPro-deprecated

Deprecated class "mf_trapezoidal_inf"

Description

Themf_trapezoidal_inf class is deprecated and will be removed in future version, useMfTrapezoidalInf instead

See Also

FisPro-deprecated

Deprecated class "mf_trapezoidal_sup"

Description

Themf_trapezoidal_sup class is deprecated and will be removed in future version, useMfTrapezoidalSup instead

See Also

FisPro-deprecated

Deprecated class "mf_triangular"

Description

Themf_triangular class is deprecated and will be removed in future version, useMfTriangular instead

See Also

FisPro-deprecated