Title: Fuzzy Logic Rule Classifier
Version: 1.0
Date: 2012-03-11
Author: Constantinos Mavridis and Ioannis N. Athanasiadis
Maintainer: Constantinos Mavridis <consmavr@gmail.com>
Description: FLR algorithm for classification
License: GPL-2 | GPL-3 [expanded from: GPL (≥ 2)]
Repository: CRAN
Date/Publication: 2014-05-06 13:19:42
Depends: combinat
Packaged: 2014-05-06 10:54:23 UTC; Sinigami
NeedsCompilation: no

Accuracy of FLR

Description

Accuracy of the flr classification.

Usage

  accIs(testData,testDataB)

Arguments

testData

an input data.frame of the test after classification.

testDataB

an input data.frame of the original test data.

Value

return the accuracy of the classification

dataset001

Description

Dataset with 296 instances if 25 attributes.

Usage

data(dataset001)

Format

A data frame with 296 instances on the following 25 variables.

state

9 US states.

county

County.

site.id

Site id.

latitude

Latitude.

longitude

Longtitude.

X2009.2011.dv

2009.2011 dv.

X2010.2012.dv

2010.2012 dv.

X2009.2011.design.value..ppm.2.3

2009-2011 design value (ppm)2,3

X2010.2012.design.value..ppm...estimated.

2010-2012 design value (ppm) [estimated].

X2009.2011.design.value.status4

2009-2011 design value status4.

percent.complete.in.20095

percent complete in 20095.

percent.complete.in.20105

percent complete in 20105.

percent.complete.in.20115

percent complete in 20115.

X2009.2011.average.percent.complete

2009-2011 average percent complete.

X..of.days.above.the.naaqs.in.2009

# of days above the naaqs in 2009.

X..of.days.above.the.naaqs.in.2010

# of days above the naaqs in 2010.

X..of.days.above.the.naaqs.in.2011

# of days above the naaqs in 2011.

X..of.days.above.the.naaqs.in.2012

# of days above the naaqs in 2012.

X4th.highest.daily.max.value.in.2009

4th highest daily max value in 2009.

X4th.highest.daily.max.value.in.2010

4th highest daily max value in 2010.

X4th.highest.daily.max.value.in.2011

4th highest daily max value in 2011.

X4th.highest.daily.max.value.in.2012.

4th highest daily max value in 2012.

column_27

Column_27.

column_29

Column_29.

class

Class category.

Source

geocommons.com

References

geocommons.com

Denormalize Fuzzy Lattices.

Description

Denormalize fuzzy lattices.

Usage

  denormDatal(fuzlat,bounds)

Arguments

fuzlat

a fuzzy lattice containing mix and max value for each instance of the data set at the first columns,from left to right, followed by className and categ.

bounds

a 2 comumn matrix containing min and max value for each instance of the dataset.

Value

return denormalized fuzzy lattice.

Constructs A Fuzzy Lattice

Description

Constructs a Fuzzy Lattice from an instance of the dataset.

Usage

  fuzzyLatticec(dF,dR,bounds)

Arguments

dF

an empty list containing just the names for each fuzzy lattice column.

dR

an instance of the dataset

bounds

a 2 comumn matrix containing min and max value for each instance of the dataset.

Value

return a fuzzy lattice (min and max value for each attribute, className,categ).

Index Calculator

Description

Returns a vector that contains the number of rules created for each class.

Usage

	indexCalc(learnedCode)

Arguments

learnedCode

a data.frame of fuzzy lattices. Each lattice is a rule created with the trainNow function.

Value

return a vector that contains the number of rules created for each class.

Graph distance matrix

Description

A matrix containing the distances of the nodes in a graph.

Usage

data(mat)

Format

A data frame of 9 rows and 9 columns.

Illinois

number

Indiana

number

Kentucky

number

Michigan

number

North.Carolina

number

Ohio

number

Pennsylvania

number

Tennessee

number

Virginia

number

Normalize Data and Denormalize data.

Description

Normalize Data to be in range of 0~1.

Usage

  normData(data1)
  denormData(data1,bounds)

Arguments

data1

an input data.frame where last instance must be the class instance and be named 'class'.

bounds

a 2 comumn matrix containing min and max value for each instance of the dataset.

Value

return normalized or denormalized data.frame.

Prepare Dataset

Description

Alters the dataset in a form that can be used for training and classification.

Usage

  prepData(data)

Arguments

data

an input data.frame where last instance must be the class instance and be named 'class'.

Value

return the data.frame without missing class instances and converts nominal attributes into numeric.

Flags Instances

Description

Randomly flags instances in order to be used as training(0) or testing(1) data with the ratio depending on variable gg.

Usage

  sepFlag(gg,data1)

Arguments

gg

percentage of instances to be used as training data for the classification.

data1

an input data.frame where last instance must be the class instance and be named 'class'.

Value

return original data with a flag column added at the end.

Creates A Boundaries File.

Description

Creates a boundaries of min and max columns for each attribute of a dataset.

Usage

  set_bounds(data1)

Arguments

data1

an input data.frame where last instance must be the class instance and be named 'class'.

Value

return a data.frame of 2 columns (min,max) for each instance of the data(NOT class).

Spatial Data Handling

Description

Creates a linear connection between spatial data in order to be used for classification.

Usage

  spatdt(data,idx,mat,pre_order=0,snd=0)
  get.cost(zzz,mat)
  get.cost2(pre_order,mat)
  get.pos(instz)
  winner.route(cost)

Arguments

data

an input data.frame

idx

indicates the position of the spatial data attribute.

mat

a matrix indicating distances

pre_order

predefined order

snd

indicates which node will be used as the starting one. The default value 0 means that the best route will be chosen, without taking into consideration which the starting node will be.

zzz

a route

instz

instance

cost

cost of routes

Value

return a list of 3 objects: a) The modified dataset, b) winner route, c) the total distance of the route.

Examples

#Import data
data(dataset001)
data<-dataset001
data(mat)

idx<-1
rhoa<-0.6
param<-"sigmoid"
pre_order<-c(1,2,3,4,5,6,7,8,9)

#Data preprocess
data<-spatdt(data,idx,mat,pre_order)

Creates Testing And Training Samples

Description

Creates testing and training samples from the original data.

Usage

  testD(data2)
  trainD(data2)

Arguments

data2

a data.frame flaged with the sepFlag function.

Value

return the training and testing samples that will be used for the classification.

Testing Phase Of FLR

Description

Implements classification using FLR on a data.frame.

Usage

  testNow(testData,learnedCode)

Arguments

testData

an input data.frame.

learnedCode

a data.frame of fuzzy lattices. Each lattice is a rule created with the trainNow function.

Value

return the testData data.frame after classification.

Training Phase Of FLR

Description

Creates rules for classification using FLR.

Usage

  trainNow(trainData,param,rhoa=0.5,l=6,x0=0.5,EPSILON=10^(-6))
  join(inpBuf,num)
  theta(x,x0,param)
  ufun(x,x0,l,param)
  valuation(fuzlat,x0,l,param)
  createNframe(trainData)
  createNlist(trainData)

Arguments

trainData

an input data.frame.

param

parameter indicating linear positive valuation for 0 and sigmoid positive valuation for 1. The default value is set to 0.

rhoa

vigilance parameter in range [0,1]. The default value is set to 0.6.

l

parameter of u and theta functions of FLR. The default value is set to 6.

x0

parameter of u and theta functions of FLR. The default value is set to 0.4.

EPSILON

parameter EPSILON.The default value is set to 10^(-6).

inpBuf

input buffer.

num

num

x

fuzzy lattice

fuzlat

fuzzy lattice

Value

return a data.frame of the learned code.