Weka BP神经网络(Neural Networks)

Weka BP神经网络(Neural Networks)分析

(2011-04-17 11:36:12)
标签:

weka

神经网络

使用

分析

分类: 算法分析(AAA)

Weka 神经网络分析

常用的神经网络就是向前反馈的 BP(Back Propagation) 网络,也叫多层前馈网络,而 BP weka 中就是由 MultilayerPerceptron 算法实现的。

(1)  Weka 主界面选择 Explorer>Open file 选择数据文件(将xls转换成csv格式)

Weka BP神经网络(Neural Networks)_第1张图片

(2)  选择 Classify 选项卡,选择 Choose 按钮,然后再选择 functions>MultiLayerPreceptron

(3)  Test Option 中选择 Use Training set ,然后单击 Choose 右侧文本框

Weka BP神经网络(Neural Networks)_第2张图片

GUI 下拉框,选择 True ,其他默认,在 Start 按钮上方选择输出层,即要预测的值,然后点击点击 OK 按钮,然后单击 Start 按钮,设置参数,

Weka BP神经网络(Neural Networks)_第3张图片

点击 Start ,运行后选择 Accept 。( momentum 带动量)

得到运行结果,在 Exploerer>Classifier output 中可以看到结果:

 

=== Run information ===

 

Scheme:       weka.classifiers.functions.MultilayerPerceptron -L 0.3 -M 0.2 -N 500 -V 0 -S 0 -E 20 -H a -G -R

Relation:     shuizhi2003-weka.filters.unsupervised.attribute.Remove-R2-7

Instances:    23

Attributes:   7

              溶解氧

              B1

              B2

              B3

              B4

              B5

              B6

Test mode:    evaluate on training data

 

=== Classifier model (full training set) ===

 

Linear Node 0

    Inputs    Weights

    Threshold    -0.7803035779165746

    Node 1    -3.5991023803526447

    Node 2     2.1968777991012516

    Node 3    3.146316946815286

Sigmoid Node 1

    Inputs    Weights

    Threshold    -2.1644039121288583

    Attrib B1    3.5316131171975034

    Attrib B2    -3.2085377318533648

    Attrib B3    -3.040826618503377

    Attrib B4    2.5982786224902887

    Attrib B5    -0.11957884133740122

    Attrib B6    1.844990948707308

Sigmoid Node 2

    Inputs    Weights

    Threshold    -1.0555748116838761

    Attrib B1    -0.3887415141267751

    Attrib B2    1.7835144295128773

    Attrib B3    -1.404906253700419

    Attrib B4    -2.458278666190213

    Attrib B5    4.483492439027104

    Attrib B6    -4.021881778908347

Sigmoid Node 3

    Inputs    Weights

    Threshold    -0.4134171735223275

    Attrib B1    0.12309277936284704

     Attrib B2    1.278817931602191

    Attrib B3    0.5473417036767106

    Attrib B4    0.18852178888400642

    Attrib B5    3.3505018112173235

    Attrib B6    2.1259296458835206

Class

    Input

    Node 0

 

 

Time taken to build model: 319.32 seconds

 

=== Evaluation on training set ===

=== Summary ===

 

Correlation coefficient                  0.9261

Mean absolute error                      0.9066

Root mean squared error                  1.1904

Relative absolute error                 34.8133 %

Root relative squared error             40.4467 %

Total Number of Instances               23   

0

你可能感兴趣的:(数据挖掘)