FCBF-java

所属分类:Java编程
开发工具:Java
文件大小:63KB
下载次数:58
上传日期:2007-09-14 16:28:07
上 传 者matriz
说明:  Feature selection is a preprocessing technique frequently used in data mining and machine learning tasks. It can reduce dimensionality, remove irrelevant data, increase learning accuracy, and improve results comprehensibility. FCBF is a fast correlation-based filter algorithm designed for high-dimensional data and has been shown effective in removing both irrelevant features and redundant features

文件列表:
FCBF-java (0, 2004-04-14)
FCBF-java\doc (0, 2004-04-14)
FCBF-java\doc\allclasses-frame.html (1129, 2004-04-14)
FCBF-java\doc\allclasses-noframe.html (1079, 2004-04-14)
FCBF-java\doc\constant-values.html (4523, 2004-04-14)
FCBF-java\doc\deprecated-list.html (4407, 2004-04-14)
FCBF-java\doc\featureselector (0, 2004-04-14)
FCBF-java\doc\featureselector\fcbf (0, 2004-04-14)
FCBF-java\doc\featureselector\fcbf\class-use (0, 2004-04-14)
FCBF-java\doc\featureselector\fcbf\class-use\DataHolder.html (6287, 2004-04-14)
FCBF-java\doc\featureselector\fcbf\class-use\MainClass.html (4922, 2004-04-14)
FCBF-java\doc\featureselector\fcbf\class-use\MetaStructure.html (4954, 2004-04-14)
FCBF-java\doc\featureselector\fcbf\class-use\ProblemAnalyzer.html (4970, 2004-04-14)
FCBF-java\doc\featureselector\fcbf\class-use\ResourceDescriptor.html (7353, 2004-04-14)
FCBF-java\doc\featureselector\fcbf\DataHolder.html (9622, 2004-04-14)
FCBF-java\doc\featureselector\fcbf\MainClass.html (9193, 2004-04-14)
FCBF-java\doc\featureselector\fcbf\MetaStructure.html (12050, 2004-04-14)
FCBF-java\doc\featureselector\fcbf\package-frame.html (1252, 2004-04-14)
FCBF-java\doc\featureselector\fcbf\package-summary.html (5963, 2004-04-14)
FCBF-java\doc\featureselector\fcbf\package-tree.html (5313, 2004-04-14)
FCBF-java\doc\featureselector\fcbf\package-use.html (5704, 2004-04-14)
FCBF-java\doc\featureselector\fcbf\ProblemAnalyzer.html (10418, 2004-04-14)
FCBF-java\doc\featureselector\fcbf\ResourceDescriptor.html (10104, 2004-04-14)
FCBF-java\doc\help-doc.html (8186, 2004-04-14)
FCBF-java\doc\index-files (0, 2004-04-14)
FCBF-java\doc\index-files\index-1.html (5333, 2004-04-14)
FCBF-java\doc\index-files\index-2.html (5467, 2004-04-14)
FCBF-java\doc\index-files\index-3.html (5557, 2004-04-14)
FCBF-java\doc\index-files\index-4.html (5533, 2004-04-14)
FCBF-java\doc\index-files\index-5.html (6398, 2004-04-14)
FCBF-java\doc\index-files\index-6.html (5566, 2004-04-14)
FCBF-java\doc\index-files\index-7.html (6410, 2004-04-14)
FCBF-java\doc\index-files\index-8.html (5713, 2004-04-14)
FCBF-java\doc\index-files\index-9.html (5272, 2004-04-14)
FCBF-java\doc\index.html (732, 2004-04-14)
FCBF-java\doc\overview-tree.html (5190, 2004-04-14)
FCBF-java\doc\package-list (22, 2004-04-14)
FCBF-java\doc\packages.html (698, 2004-04-14)
FCBF-java\doc\stylesheet.css (1328, 2004-04-14)
FCBF-java\fcbf_1_0.jar (17408, 2004-04-14)
... ...

************************************************************************ README FILE prepared by Ravi Bhimavarapu and Huan Liu on April 14, 2004 at ASU Java implementation of FCBF This software is provided free of charge to the research community as an academic software package with no commitment in terms of support or maintenance. The algorithm FCBF and its detailed description can be found in ICML 2003 "Feature Selection for High-Dimensional Data: A Fast Correlation-Based Filter Solution" by Lei Yu and Huan Liu. This readme file consists of five sections as follows: Section 0 – Contents of the zip file. Section A – Run FCBF with the package Section B – Format of the input Section C – Use the classes in the package Section D – Miscellaneous Information Prepared and implemented by Ravi Sekhar Reddy Bhimavarapu for the data mining and machine learning lab at ASU 4/14/04 Ravi can be reached at ravi.bhim@asu.edu. *********************************************************************** Section 0 TABLE OF CONTENTS After inflating the zip file these are the files and directories you would find. * ‘fcbf_1_0.jar’ package file. * ‘doc’ directory which has the documentation. * ‘test’ directory which the sample test data (‘happy’ and ‘lc’). * This README document. Section A How TO RUN 1) Go the directory which has the jar file 2) Inflate the ‘jar’ file: run `jar xvf fcbf_1_0.jar' at command line 3) Run “java featureselector.fcbf.MainClass ‘stem’ ” NOTE: The ‘delta’ value is optional. If no value is specified, the default value for ‘delta’ is 0. Two simple datasets "happy" and lung-cancer dataset "lc" are provided in the "test" directory. Command to run on data "lc" is For windows, ‘java featureselector.fcbf.MainClass test\lc 0.13’ ( delta is 0.13 here ) For unix/linux, ‘java featureselector.fcbf.MainClass test/lc 0.13’ ( delta is 0.13 here ) Section B DATA FORMATS The input is specified in two files, one containing the data (with .data extension) and the other containing the metadata information( with .rc extension). Both files should have the same stem, e.g., happy.data and happy.rc and the stem here is ‘happy’. The .data file -------------------- The data file contains one instance on each line and the attribute values are comma separated. The class label is given at the end of each instance. The .rc file ------------------ As mentioned above, the rc file contains information about the metadata. The format is explained below using the "happy.rc" file as an example. 2,unhappy,happy. # Number of classes, class_label1,class_label2,.... 3 # Number of attributes. age - twenties,fifties. # Attribute 1 label - Attribute values.... height - short,tall. coll - not_college,college. Section C Importing the classes from the package in order to use them in your implementation. The classes are present in the package “featureselector.fcbf”, so for importing the classes you need to add the line in the import declarations import featureselector.fcbf.*; The source files for the 5 classes in the package are also compressed in the ‘jar’ file. To get an idea of how the classes should be instantiated and called, please look at the source file MainClass.java . One could also use the MainClass class to run the FCBF on a specified data set (steps shown in Section A) Section D Documentation for the classes – The documentation for the classes can be found in the directory ‘doc’

近期下载者

相关文件


收藏者