转：matlab+spider+weka

 

转自数据挖掘青年　http://blogger.org.cn/blog/more.asp?name=DMman&id=27375

一 spider主页http://www.kyb.mpg.de/bs/people/spider/ （也可以在google上搜索spider matlab得到），关于它的介绍可以参考网址资料

二 使用时为matlab+spider+Weka；因为spider中的一些算法引用了Weka，比如j48

安装注意：

1 matlab7（R14）

6.5版本对java的支持不够，还没有开发javaclasspath等函数

??? Undefined function or variable 'javaclasspath'.
??? Undefined function or variable 'javaaddclasspath'.

2 jre1.4.2

matlab7自带的是1.4.2；matlab6自带的是1.3.可以在D:\MATLAB7\sys\java\jre\win32下看到。如果装了matlab7，使用它自带的1.4.2就可以了，尤其不要使用1.6，因为1.6太新了，matlab还不支持。可以在Matlab下使用 version -java查看JVM版本。

如果你想使用1.5的话，C:\Program Files\Java\jre1.5.0_10；把jre1.5.0_10这个文件夹拷贝到D:\MATLAB7\sys\java\jre\win32下，然后增加环境变量MATLAB_JAVA：D:\MATLAB7\sys\java\jre\win32\jre1.5.0_10。这一步如果有问题的话，重启Matlab会给出错误提示。找不到什么什么文件...

3 Weka3.4.10

使用weka版本低一些即可，高的不行，因为高版本的weka可能是用高版本的jvm支持的。

我使用的组合是 matlab7（R14）+jre1.4.2（matlab7自带的，不需要任何设置）+Weka3.4.10

weka 下载：http://www.cs.waikato.ac.nz/ml/weka/

三 使用方法

1 下载spider，有core和extra两个压缩包，把他们解压到同一个文件夹spider下面，然后放到$matlabroot\toolbox下面

2下载weka3.4.10，找到weka.jar放到$matlabroot\java\jar下面

3 启动Matlab打开$matlabroot\toolbox\spider\use_spider.m运行

提示spider的一些信息和 WEKA support enabled!表示成功了。

然后可以使用 help spider命令查看信息，他的功能列出如附录，然后就可以训练了。

四 一个简单的例子

X=rand(50)-0.5; Y=sign(sum(X,2));
dtrain=data(X,Y);
%生成训练集，也可以使用load()从文件读取

model=train(svm,dtrain));
%使用函数train（）训练模型

rtest=test(dtest,model);
%使用训练好的模型对验证集dtest测试，返回测试结果

五 附录spider信息

最新spider Version 1.71 (24/7/2006)

Basic library objects. 
data - Storing input data and output results 
data_global - Implementation of data object that limits memory overhead
algorithm - Generic algorithm object
group - Groups sets of objects together (algorithms or data) 
loss - Evaluates loss functions
get_mean - Takes mean loss over groups of algs
chain - Builds chains of objects: output of one to input of another
param - To train and test different hyperparameters of an object
cv - Cross validation using objects given data
kernel - Evaluates and caches kernel functions
distance - Evaluates and caches distance functions

Statistical Tests objects.
wilcoxon - Wilcoxon test of statistical significance of results
corrt_test - Corrected resampled t-test - for dependent trials

Dataset objects.
spiral - Spiral dataset generator.
toy - Generator of dataset with only a few relevant features
toy2d - Simple 2d Gaussian problem generator
toyreg - Linear Regression with o outputs and n inputs 

Pre-Processing objects
normalize - Simple normalization of data
map - General user specified mapping function of data

Density Estimation objects.
parzen - Parzen's windows kernel density estimator
indep - Density estimator which assumes feature independence
bayes - Classifer based on density estimation for each class
gauss - Normal distribution density estimator

Pattern Recognition objects.
svm - Support Vector Machine (svm)
c45 - C4.5 for binary or multi-class 
knn - k-nearest neighbours
platt - Conditional Probability estimation for margin classifiers
mksvm - Multi-Kernel LP-SVM
anorm - Minimize the a-norm in alpha space using kernels
lgcz - Local and Global Consistent Learner 
bagging - Bagging Classifier
adaboost - ADABoost method
hmm - Hidden Markov Model 
loom - Leave One Out Machine 
l1 - Minimize l1 norm of w for a linear separator 
kde - Kernel Dependency Estimation: general input/output machine
dualperceptron - Kernel Perceptron
ord_reg_perceptron - Ordinal Regression Perceptron (Shen et al.)
splitting_perceptron - Splitting Perceptron (Shen et al.)
budget_perceptron - Sparse, online Pereceptron (Crammer et al.)
randomforest - Random Forest Decision Trees WEKA-Required
j48 - J48 Decision Trees for binary WEKA-Required

Multi-Class and Multi-label objects. 
one_vs_rest - Voting method of one against the rest (also for multi-label)
one_vs_one - Voting method of one against one
mc_svm - Multi-class Support Vector Machine by J.Weston
c45 - C4.5 for binary or multi-class 
knn - k-nearest neighbours

Feature Selection objects.
feat_sel - Generic object for feature selection + classifier
r2w2_sel - SVM Bound-based feature selection
rfe - Recursive Feature Elimination (also for the non-linear case)
l0 - Dual zero-norm minimization (Weston, Elisseeff)
fsv - Primal zero-norm based feature selection (Mangasarian)
fisher - Fisher criterion feature selection
mars - selection algorithm of Friedman (greedy selection)
clustub - Multi-class feature selection using spectral clustering
mutinf - Mutual Information for feature selection.

Regression objects.
svr - Support Vector Regression
gproc - Gaussian Process Regression 
relvm_r - Relevance vector machine 
multi_rr - (possibly multi-dimensional) ridge regression 
mrs - Multivariate Regression via Stiefel Constraints 
knn - k-nearest neighbours
multi_reg - meta method for independent multiple output regression
kmp - kernel matching pursuit
kpls - kernel partial least squares
lms - least mean squared regression [now obselete due to multi_rr]
rbfnet - Radial Basis Function Network (with moving centers)
reptree - Reduced Error Pruning Tree WEKA-Required

Model Selection objects.
gridsel - select parameters from a grid of values 
r2w2_sel - Selecting SVM parameters by generalization bound
bayessel - Bayessian parameter selection 

Unsupervised objects.
one_class_svm - One class SVM
kmeans - K means clustering
kvq - Kernel Vector Quantization
kpca - Kernel Principal Components Analysis
ppca - Probabilistic Principal Component Analysis
nmf - Non-negative Matrix factorization
spectral - Spectral clustering
mrank - Manifold ranking
ppca - Probabilistic PCA

Reduced Set and Pre-Image objects.
pmg_mds - Calculate Pre-Images based on multi-dimensional scaling
pmg_rr - Calculate Pre-Images based on learning and ridge regression
rsc_burges - Bottom Up Reduced Set; calculates reduced set based on gradient descent
rsc_fp - Bottom Up Reduced Set; calculates reduced set for rbf with fixed-point iteration schemes
rsc_mds - Top Down Reduced Set; calculates reduced set with multi-dimensional scaling
rsc_learn - Top Down Reduced Set; calculates reduced set with ridge regression
rss_l1 - Reduced Set Selection via L1 penalization
rss_l0 - Reduced Set Selection via L0 penalization
rss_mp - Reduced Set Selection via matching pursuit