The arboretum procedure



Yüklə 3.07 Mb.

səhifə103/148
tarix30.04.2018
ölçüsü3.07 Mb.
1   ...   99   100   101   102   103   104   105   106   ...   148

Notice that the MLP network with 3 hidden units correctly classifies all cases in the training data set.

                    MLP with 3 Hidden Units

                   Fits Statistics for the Training Data Set

               Train:        Train:         Train:       Train: Root

              Akaike's       Average       Maximum          Final

            Information      Squared       Absolute      Prediction

             Criterion.      Error.         Error.         Error.

              34.0000      3.7039E-14    .0000030981    .00000020177

 

                  Train:           Train: Number



            Misclassification        of Wrong

                  Rate.          Classifications.

                    0                    0     

PROC FREQ Misclassification Table for the Training Data

                       MLP with 3 Hidden Units

                            Misclassification Table

                              TABLE OF F_C BY I_C

                F_C(From: C)     I_C(Into: C)

                Frequency    |

                Percent      |

                Row Pct      |

                Col Pct      |       1|       2|       3|  Total

                             |        |        |        |

                -------------+--------+--------+--------+

                           1 |      8 |      0 |      0 |      8

                             |   4.44 |   0.00 |   0.00 |   4.44

                             | 100.00 |   0.00 |   0.00 |

                             | 100.00 |   0.00 |   0.00 |

                -------------+--------+--------+--------+

                           2 |      0 |     59 |      0 |     59

                             |   0.00 |  32.78 |   0.00 |  32.78

                             |   0.00 | 100.00 |   0.00 |

                             |   0.00 | 100.00 |   0.00 |

                -------------+--------+--------+--------+

                           3 |      0 |      0 |    113 |    113

                             |   0.00 |   0.00 |  62.78 |  62.78

                             |   0.00 |   0.00 | 100.00 |

                             |   0.00 |   0.00 | 100.00 |

                -------------+--------+--------+--------+

                Total               8       59      113      180

                                 4.44    32.78    62.78   100.00



PROC GPLOT Plot of the Classification Results


PROC GGONTOUR Plots of the Posterior Probabilities

The legend at the bottom of the chart identifies the target level.






Copyright 2000 by SAS Institute Inc., Cary, NC, USA. All rights reserved.


 

PROC GPLOT creates a scatter plot of the Rings training data.

proc gplot data=sampsio.dmdring;

   plot y*x=c /haxis=axis1 vaxis=axis2;

   symbol c=black i=none v=dot;

   symbol2 c=red i=none v=square;

   symbol3 c=green i=none v=triangle;

   axis1 c=black width=2.5  order=(0 to 30 by 5);

   axis2 c=black width=2.5 minor=none order=(0 to 20 by 2);

   title 'Plot of the Rings Training Data';

run;



 

The PROC NEURAL statement invokes the procedure. The DATA= option identifies

the training data set that is used to fit the model. The DMDBCAT= option identifies

the DMDB catalog. The RANDOM= option specifies the random number seed.

proc neural data=sampsio.dmdring 

                         dmdbcat=sampsio.dmdring 

                         random=789; 



 

The INPUT statement specifies an  interval input layer. The LEVEL= option

specifies the measurement level. The ID= option specifies an identifier for

the interval input layer.

   input x y / level=interval id=i;



 

The TARGET statement defines an output layer. The output layer computes

predicted values and compares those predicted values with the value of the

target variable. The ID= option specifies an identifier for the output layer.

The LEVEL= option specifies the target measurement level.   By default, for

nominal targets the combination function is set to linear,  the activation

function is set to mlogistic, and the error function is set to mbernoulli.

target c / id=o level=nominal;




 

The HIDDEN statement defines the number of hidden units  that are used

to perform the internal computations. By default, the input units are connected

to each hidden unit and each  hidden unit is connected to the output unit.

The ID= option specifies an identifier for the hidden unit.

  

   hidden 3 / id=h;




 

The PRELIM statement causes the procedure to search for the best starting

weights for subsequent training. The integer value of  5 specifies to use

5 preliminary runs. The weights from the seed with the smallest objective

function among all runs is chosen. Preliminary training may help prevent the

network from converging in a local minima. 

    prelim 5;





Dostları ilə paylaş:
1   ...   99   100   101   102   103   104   105   106   ...   148


Verilənlər bazası müəlliflik hüququ ilə müdafiə olunur ©genderi.org 2017
rəhbərliyinə müraciət

    Ana səhifə