Escolar Documentos
Profissional Documentos
Cultura Documentos
Bernhard Pfahringer
(based on material by Eibe Frank, Mark Hall, and Peter Reutemann)
Department of Computer Science, University of Waikato, New Zealand
Classification and Regression Clustering Association Rules Attribute Selection Data Visualization
Machine learning/data mining software written in Java (distributed under the GNU Public License) Used for research, education, and applications Complements Data Mining by Witten & Frank Main features:
Comprehensive set of data pre-processing tools, learning algorithms and evaluation methods Graphical user interfaces (incl. data visualization) Environment for comparing learning algorithms
5/2/2012
University of Waikato
History
Develop state-of-the art workbench of data mining tools Explore fielded applications Develop new fundamental methods
5/2/2012
University of Waikato
History (2)
Late 1992 - funding was applied for by Ian Witten 1993 - development of the interface and infrastructure WEKA acronym coined by Geoff Holmes WEKAs file format ARFF was created by Andrew Donkin ARFF was rumored to stand for Andrews Ridiculous File Format Sometime in 1994 - first internal release of WEKA TCL/TK user interface + learning algorithms written mostly in C Very much beta software Changes for the b1 release included (among others):
Ambiguous and Unsupported menu commands removed. Crashing processes handled (in most cases :-) October 1996 - first public release: WEKA 2.1
5/2/2012
University of Waikato
History (3)
July 1997 - WEKA 2.2 Schemes: 1R, T2, K*, M5, M5Class, IB1-4, FOIL, PEBLS, support for C5 Included a facility (based on Unix makefiles) for configuring and running large scale experiments Early 1997 - decision was made to rewrite WEKA in Java Originated from code written by Eibe Frank for his PhD Originally codenamed JAWS (JAva Weka System) May 1998 - WEKA 2.3 Last release of the TCL/TK-based system Mid 1999 - WEKA 3 (100% Java) released Version to complement the Data Mining book Development version (including GUI)
5/2/2012
University of Waikato
WEKA: versions
5/2/2012
University of Waikato
java weka.gui.GUIChooser
5/2/2012
University of Waikato
11
5/2/2012
University of Waikato
12
5/2/2012
University of Waikato
13
5/2/2012
University of Waikato
14
Data can be imported from a file in various formats: ARFF, CSV, C4.5, binary Data can also be read from a URL or from an SQL database (using JDBC) Pre-processing tools in WEKA are called filters WEKA contains filters for:
5/2/2012
University of Waikato
15
5/2/2012
University of Waikato
16
5/2/2012
University of Waikato
17
5/2/2012
University of Waikato
18
5/2/2012
University of Waikato
19
5/2/2012
University of Waikato
20
5/2/2012
University of Waikato
21
5/2/2012
University of Waikato
22
5/2/2012
University of Waikato
23
5/2/2012
University of Waikato
24
5/2/2012
University of Waikato
25
5/2/2012
University of Waikato
26
5/2/2012
University of Waikato
27
5/2/2012
University of Waikato
28
5/2/2012
University of Waikato
29
5/2/2012
University of Waikato
30
5/2/2012
University of Waikato
31
5/2/2012
University of Waikato
32
5/2/2012
University of Waikato
33
5/2/2012
University of Waikato
34
5/2/2012
University of Waikato
35
5/2/2012
University of Waikato
36
Classifiers in WEKA are models for predicting nominal or numeric quantities Implemented learning schemes include:
Decision trees and lists, instance-based classifiers, support vector machines, multi-layer perceptrons, logistic regression, Bayes nets,
Meta-classifiers include:
5/2/2012
5/2/2012
University of Waikato
38
5/2/2012
University of Waikato
39
5/2/2012
University of Waikato
40
5/2/2012
University of Waikato
41
5/2/2012
University of Waikato
42
5/2/2012
University of Waikato
43
5/2/2012
University of Waikato
44
5/2/2012
University of Waikato
45
5/2/2012
University of Waikato
46
5/2/2012
University of Waikato
47
5/2/2012
University of Waikato
48
5/2/2012
University of Waikato
49
5/2/2012
University of Waikato
50
5/2/2012
University of Waikato
51
5/2/2012
University of Waikato
52
5/2/2012
University of Waikato
53
5/2/2012
University of Waikato
54
5/2/2012
University of Waikato
55
5/2/2012
University of Waikato
56
5/2/2012
University of Waikato
57
5/2/2012
University of Waikato
58
5/2/2012
University of Waikato
59
5/2/2012
University of Waikato
60
5/2/2012
University of Waikato
61
5/2/2012
University of Waikato
62
5/2/2012
University of Waikato
63
5/2/2012
University of Waikato
64
5/2/2012
University of Waikato
65
5/2/2012
University of Waikato
66
5/2/2012
University of Waikato
67
5/2/2012
University of Waikato
68
5/2/2012
University of Waikato
69
5/2/2012
University of Waikato
70
5/2/2012
University of Waikato
71
QuickTime and a T IFF (LZW) decompressor are needed to see this picture.
5/2/2012
University of Waikato
72
QuickTime and a T IFF (LZW) decompressor are needed to see this picture.
5/2/2012
University of Waikato
73
5/2/2012
University of Waikato
74
5/2/2012
University of Waikato
75
5/2/2012
University of Waikato
76
5/2/2012
University of Waikato
77
5/2/2012
University of Waikato
78
5/2/2012
University of Waikato
79
5/2/2012
University of Waikato
80
5/2/2012
University of Waikato
81
5/2/2012
University of Waikato
82
5/2/2012
University of Waikato
83
5/2/2012
University of Waikato
84
5/2/2012
University of Waikato
85
5/2/2012
University of Waikato
86
5/2/2012
University of Waikato
87
5/2/2012
University of Waikato
88
5/2/2012
University of Waikato
89
5/2/2012
University of Waikato
90
5/2/2012
University of Waikato
91
5/2/2012
University of Waikato
92
WEKA contains clusterers for finding groups of similar instances in a dataset Some implemented schemes are:
Clusters can be visualized and compared to true clusters (if given) Evaluation based on loglikelihood if clustering scheme produces a probability distribution
5/2/2012
University of Waikato
93
5/2/2012
University of Waikato
94
5/2/2012
University of Waikato
95
5/2/2012
University of Waikato
96
5/2/2012
University of Waikato
97
5/2/2012
University of Waikato
98
5/2/2012
University of Waikato
99
5/2/2012
University of Waikato
100
5/2/2012
University of Waikato
101
5/2/2012
University of Waikato
102
WEKA contains the Apriori algorithm (among others) for learning association rules
milk, butter bread, eggs (with confidence 0.9 and support 2000)
Apriori can compute all rules that have a given minimum support and exceed a given confidence
University of Waikato 103
5/2/2012
5/2/2012
University of Waikato
104
5/2/2012
University of Waikato
105
5/2/2012
University of Waikato
106
Panel that can be used to investigate which (subsets of) attributes are the most predictive ones Attribute selection methods contain two parts:
A search method: best-first, forward selection, random, exhaustive, genetic algorithm, ranking An evaluation method: correlation-based, wrapper, information gain, chi-squared,
5/2/2012
5/2/2012
University of Waikato
108
5/2/2012
University of Waikato
109
5/2/2012
University of Waikato
110
5/2/2012
University of Waikato
111
5/2/2012
University of Waikato
112
5/2/2012
University of Waikato
113
Visualization very useful in practice: e.g. helps to determine difficulty of the learning problem WEKA can visualize single attributes (1-d) and pairs of attributes (2-d)
Color-coded class values Jitter option to deal with nominal attributes (and to detect hidden data points) Zoom-in function
University of Waikato 114
5/2/2012
5/2/2012
University of Waikato
115
5/2/2012
University of Waikato
116
5/2/2012
University of Waikato
117
5/2/2012
University of Waikato
118
5/2/2012
University of Waikato
119
5/2/2012
University of Waikato
120
5/2/2012
University of Waikato
121
5/2/2012
University of Waikato
122
5/2/2012
University of Waikato
123
5/2/2012
University of Waikato
124
Performing experiments
Experimenter makes it easy to compare the performance of different learning schemes For classification and regression problems Results can be written into file or database Evaluation options: cross-validation, learning curve, hold-out Can also iterate over different parameter settings Significance-testing built in!
University of Waikato 125
5/2/2012
5/2/2012
University of Waikato
126
5/2/2012
University of Waikato
127
5/2/2012
University of Waikato
128
5/2/2012
University of Waikato
129
5/2/2012
University of Waikato
130
5/2/2012
University of Waikato
131
5/2/2012
University of Waikato
132
5/2/2012
University of Waikato
133
5/2/2012
University of Waikato
134
5/2/2012
University of Waikato
135
Java-Beans-based interface for setting up and running machine learning experiments Data sources, classifiers, etc. are beans and can be connected graphically Data flows through components: e.g., data source -> filter -> classifier -> evaluator Layouts can be saved and loaded again later cf. Clementine
University of Waikato 136
5/2/2012
5/2/2012
University of Waikato
137
5/2/2012
University of Waikato
138
5/2/2012
University of Waikato
139
5/2/2012
University of Waikato
140
5/2/2012
University of Waikato
141
5/2/2012
University of Waikato
142
5/2/2012
University of Waikato
143
5/2/2012
University of Waikato
144
5/2/2012
University of Waikato
145
5/2/2012
University of Waikato
146
5/2/2012
University of Waikato
147
5/2/2012
University of Waikato
148
5/2/2012
University of Waikato
149
5/2/2012
University of Waikato
150
5/2/2012
University of Waikato
151
Sourceforge.net Downloads
5/2/2012
University of Waikato
152
5/2/2012
University of Waikato
153
45 projects currently (30/01/07) listed on the WekaWiki Incorporate/wrap WEKA GRB Tool Shed - a tool to aid gamma ray burst research YALE - facility for large scale ML experiments GATE - NLP workbench with a WEKA interface Judge - document clustering and classification RWeka - an R interface to Weka Extend/modify WEKA BioWeka - extension library for knowledge discovery in biology WekaMetal - meta learning extension to WEKA Weka-Parallel - parallel processing for WEKA Grid Weka - grid computing using WEKA Weka-CG - computational genetics tool library
5/2/2012
University of Waikato
154
Pentaho The leader in Open Source Business Intelligence (BI) September 2006 Pentaho acquires the Weka project (exclusive license and SF.net page) Weka will be used/integrated as data mining component in their BI suite Weka will be still available as GPL open source software Most likely to evolve 2 editions:
5/2/2012
University of Waikato
155
Limitations of WEKA
Traditional algorithms need to have all data in main memory ==> big datasets are an issue Solution:
Incremental schemes Stream algorithms MOA Massive Online Analysis (not only a flightless bird, but also extinct!)
5/2/2012
University of Waikato
156
WEKA is available at http://www.cs.waikato.ac.nz/ml/weka Also has a list of projects based on WEKA (probably incomplete list of) WEKA contributors:
Abdelaziz Mahoui, Alexander K. Seewald, Ashraf M. Kibriya, Bernhard Pfahringer, Brent Martin, Peter Flach, Eibe Frank, Gabi Schmidberger, Ian H. Witten, J. Lindgren, Janice Boughton, Jason Wells, Len Trigg, Lucio de Souza Coelho, Malcolm Ware, Mark Hall, Remco Bouckaert, Richard Kirkby, Shane Butler, Shane Legg, Stuart Inglis, Sylvain Roy, Tony Voyle, Xin Xu, Yong Wang, Zhihai Wang
5/2/2012
University of Waikato
157