Hybrid Fuzzy Rule Based Classification Algorithm

The purpose of this document is to design a strategy for hybrid fuzzy rule base classification algorithm using the weka tool. This document outlines the functional requirements for hybrid fuzzy rule based classification algorithm. This document discusses the project’s goals and parameters, while giving descriptions about the potential design issues. The requirements are specified according to the finished product.

1.2 Document Conventions
This document has been written on the following style Font style Headings Sub – Headings Data Line Spacing Times New Roman 16 Bold 14 Bold 12 Regular 1.5 Lines

1.3 Intended Audience and Reading Suggestions
The document will have a wide application in which data mining solutions is required. Besides, researchers who are interested in the field of Data mining will find this system as a useful tool.

1.4 Project scope
The purpose of the project is to hybridize the underlying concepts of fuzzy rule base classification algorithm to deal with additional aspects of data imperfection. Objective of the project is to integrate different models of fuzzy rules-based classification algorithm so as to bring out a new algorithm. Goal of the project is to produce better accuracy than the other models being used for classification.

CHAPTER II

2. Overall Description
2.1 Product Perspective
Classification is a data mining algorithm that creates a step-by-step guide for how to determine the output of a new data instance. There are various models available for doing classification like FuzzyNN, FuzzyRoughNN,NN, and FuzzyOwnershipNN etc. In this research work an attempt is made to integrate these models to bring out a more efficient model that can inherit the properties of these models and provide better performance in classification process.

2.2 Product Features
The system has four modules. o o o o Creating an Instance to load input data source. Assigning Classes to rank the attribute. Applying Classification algorithm to analyze result set. Visualizing the result set.

2.3 User Classes and Characteristics
The various user classes of this project are students, research scholars and people interested in area of Data mining.

2.4 Operating Environment
Since Weka is written in pure java, it is able to run on most hardware and operating system. Naturally, the hardware requirements will vary depending on nature and size of problem, though the basic requirements to run Weka are quite small and should be abl e to run on even older hardware. Windows Environment is used for this work. 2.4.1 Hardware Specification Processor Memory Hard Disk Drive : Intel Pentium IV : 512 MB RAM : 2 GB

2.4.2 Software Specification Operating System Front End Different Data Set : Windows 7 : Weka : Machine Learning Repository

2.5 Design and Implementation Constraints
System Constraints o o o o A Java2 compliant Virtual Machine. The Java JRE / JDK. A graphics capable machine. Weka 3.7.2 version. Netbeans 6.9 version

Processing Constraints The system needs good speed i.e. a good configuration processor and RAM, since datasets of different sizes are used.

2.6 User Documentation
This product uses WEKA, which has a user manual. Only the work and modifications that has been done to weka has to be added. Manual will include product overview, complete configuration of the used system, technical details, and contact information.

2.7 Assumptions and Dependencies
Front-end (user interaction) Weka (Waikato Environment for Knowledge Analysis) is a popular suite of machine learning software written in Java, developed at the University of Waikato, New Zealand. Weka is free software available under the GNU General Public License. Dependencies: Weka and different Algorithm/methods used in Weka. specific hardware and Software constraints. No

CHAPTER III

3. System Features
This section describes in detail the various features of this system. They are listed below.

3.1 Creating an Instance
This feature allows the user to select a dataset from different data sources. 3.1.1 Stimulus/Response Sequences Data set is preprocessed in order to apply classifier. 3.1.2 Functional Requirements Purpose Inputs Processing Outputs Processing of Dataset. Dataset. LoadingdatasetAssigningclasscrossvalidationbatch classifierclassifier performance EvaluatorText viewer. Loading dataset for processing.

3.2 Ranking Attributes
This feature ranks the attribute based on method that is used to intersect.

3.2.1 Stimulus/ Response Sequences Attribute is ranked based on the method that is been intersected for ranking the attributes 6

3.2.2 Functional Requirements Purpose Inputs Processing Outputs To rank the attribute Preprocessed Dataset Ranking of attribute based on fuzzy algorithm Attributes are ranked

3.3 Performing Classification
This feature selects the classification for the ranked dataset. 3.3.1 Stimulus/Response Sequences Selecting fuzzy hybrid classification algorithm for classification. 3.3.2 Functional Requirements Purpose Inputs Processing Outputs To select algorithm for classification Ranked Dataset. Classification Process. Classification algorithm is selected

3.4 Visualizing Result
Classification is performed on the dataset. 3.4.1 Stimulus/Response Sequences To perform classification by the classification algorithm that is been selected on the dataset.

3.4.2 Functional Requirements Purpose Inputs Processing Outputs To calculate the accuracy of the selected dataset. Processed Data set The classification is done on the dataset Correctly classified instances are available.

Chapter IV

External Interface Requirements

4.1 User Interfaces Weka Software is used and hence to implement this work, the weka software is modified. The user interfaces corresponds to interfaces of Weka. 4.2 Hardware Interfaces Since Weka is written in pure java, it is able to run on most hardware and operating systems. Naturally, the hardware requirements will vary depending on the nature and size of problem, though the basic requirements to run Weka are quite small and should be able to run on even older hardware. For the purposes of this project, we will be using the Weka System in a windows 4.3 Software Interfaces o o Java JRE/JDK Weka 3.7.2 version used here environment.

CHAPTER V

5. Other Non-Functional Requirements
5.1   RAM 5.2 Performance Requirements To execute the use case of system the following requirements are needed Weka Netbeans problem may arise when large dataset is applied.

Safety Requirements The Result set can be stored and taken as backup which can be used for future reference. 5.3 Software Quality Attributes : Measures will be taken so as to work with open : The dataset are preprocessed and it can be reusable in any algorithm. Reliability : The system uses mathematical operations on data set, hence the system is accurate and reliable. source system.

Adaptability Reusability

CHAPTER VI 6. Other Requirements
6.1 Steps to import Weka code into Netbeans: Manual checkout: Requirements   

ANT Java 1.4 or later for Weka 3.7.x Java 1.5 or later for Weka 3.5.x or later

Extract the source code Extract the source code from the weka-src.jar with any archive manager that can handle the ZIP file format into a temporary directory (don’t forget to re-recreate the folder structure when extracting). Step 1 : Create a new Java application File -> New Project

Step 2: Enter the project details

Project name: weka Project location: /tmp Create main class: book version: weka.gui.GUIChoose developer version: weka.gui.Main

Step 3: Click on Finish Move sources Move the weka directory from the previously extracted weka-src.jar into the following directory (overwrite existing files and directories):/tmp/weka/src Go back into Netbeans, it should automatically notice that there are new files Build the project Build -> Build Main Project Run Weka with Run -> Run Main Project

Appendix A

Analysis Model

Applying the Hybrid algorithm for classification

Fig : Architectural diagram showing the flow of the system

References

 1. D. Aha, Instance-based learning algorithm, Machine Learning, vol. 6, pp. 37{66, 1991. 2. R.B. Bhatt and M.Gopal, FRID: Fuzzy-Rough Interactive Dichotomizers, IEEE International Conference on Fuzzy Systems (FUZZ-IEEE’04), pp. 1337{1342, 2004. 3. H. Bian and L.
Mazlack, Fuzzy-Rough Nearest-Neighbor Classication Approach, Proceeding of the 22nd International Conference of the North American Fuzzy Information Processing Society (NAFIPS), pp. 500{505, 2003. 4. C.L. Blake and C.J. Merz. UCI Repository of machine learning databases. Irvine, University of California, 1998. http://archive.ics.uci.edu/ml/ 5. W.W. Cohen, Fast e ective rule induction, In Machine Learning: Proceedings of the 12th International Conference, pp. 115.