The Evolutionary Pre-Processor (version 3.0)
Batch Run Report

 

Batch Details
Batch run name german_1
Description German 1 on Dobson
This report file results/final/german_1/german_1_report.html
data file /home/users/jsherrah/EPrep3/data/german_1.dat
Time of completion Sun Jun 21 03:51:43 1998
Duration of Batch 11 hours, 38 minutes, 28 seconds
Random Seed 571123881 (from clock)
Average generations per run 42.50
Average failed feature creations per run 6827.60
Average fitness evalutions per run 144253.80
Test Set Improvement 9.60 %
 
 

Data Partition
 
Class Training Validation Test Total
0 350 175 175 700
1 150 75 75 300
Total 500 250 250 1000

Summary of Results

  Original Classification Errors (%)
Classifier Training Validation Test
Maximum Likelihood(ML) 22.00 36.00 36.40
Min. Distance to Means(MDTM) 37.00 36.40 40.80

  EPrep Best-of-run Classification Errors (%)
Run
Training
Validation
Test
McNemar confidence
# Features
# Inputs
# Nodes
Classifier
1 22.80 24.80 26.80 0.995 11 10 54 ML
2 27.60 26.80 28.40 0.966 3 12 169 MDTM
3 17.60 28.40 32.80 0.846 18 19 825 ML
4 27.40 28.80 31.20 0.889 9 20 929 MDTM
5 22.00 29.20 27.60 0.999 14 13 20 ML
6 23.20 26.00 33.20 0.819 14 15 83 ML
7 21.60 28.40 29.20 0.974 12 18 251 ML
8 16.40 30.00 34.00 0.788 18 20 355 ML
9 18.60 29.60 27.20 0.998 18 19 770 ML
10 23.20 27.20 26.40 1.000 14 16 70 ML
Ave. 22.04 (3.56 ) 27.92 (1.59 ) 29.68 (2.73 ) 0.93 (0.08 ) 13.10 (4.46 ) 16.20 (3.40 ) 352.60(335.56) ML

  Confusion matrix for Best Ever Individual from run 1
Class
Predicted
Total
1
2
Ground Truth
1
147 28 175
2
39 36 75
Total
186 64 250

Average Operator Probabilities
Operator Average Probability
Delete-Feature Mutation 0.093
Add-Feature Mutation 0.110
Hoist Mutation 0.102
Truncate Mutation 0.096
Swap Mutation 0.103
One-Symbol Mutation 0.099
All-Nodes Mutation 0.098
One-Node Mutation 0.093
Grow Mutation 0.097
High-Level Crossover 0.108
 

Number of Run Terminations attributed to each Criterion
Termination Criterion Number of Terminations
TP Criterion 0
GL Criterion 5
Max. Generations 5
Client Abort 0
Zero Validation Error 0
Total 10
 
 

Related Data Files
Description Filename
german_1_gen.m Matlab plot generation function
german_1_bogf_ave.{eps,gif} Best-of-generation Fitness, averaged over 10 runs
german_1_bogv_ave.{eps,gif} Best-of-generation Validation Set Error, averaged over 10 runs
german_1_avef_ave.{eps,gif} Average fitness, averaged over 10 runs
german_1_stdf_ave.{eps,gif} Standard deviation of fitness, averaged over 10 runs
german_1_nftr_ave.{eps,gif} Average number of features per individual, averaged over 10 runs
german_1_nnode_ave.{eps,gif} Average number of nodes per individual, averaged over 10 runs
german_1_nint_ave.{eps,gif} Average number of introns per individual, averaged over 10 runs
german_1_ntrl_ave.{eps,gif} Average number of RAT trials per individual, averaged over 10 runs
german_1_optimp_ave.{eps,gif} Average improvement in fitness due to optimisation, averaged over 10 runs
german_1_opprob_ave.{eps,gif} Average probability of each genetic operator, averaged over 10 runs
german_1_run_x.dat Binary data file containing results of run x (read by Matlab functions)
german_1_bor_x.prep 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Best-of-run individual for run x
german_1_run_x.corr 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Feature correlation file for run x
german_1_tst_bor_x.pred 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Test set predictions for best-of-run individual from run x
german_1_bogf_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Best-of-generation Fitness for run x
german_1_bogv_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Best-of-generation Validation Set Error for run x
german_1_avef_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Average fitness for run x
german_1_stdf_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Standard deviation of fitness for run x
german_1_nftr_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Average number of features per individual for run x
german_1_nnode_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Average number of nodes per individual for run x
german_1_nint_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Average number of introns per individual for run x
german_1_ntrl_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Average number of RAT trials per individual for run x
german_1_optimp_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Average improvement in fitness due to optimisation for run x
german_1_opprob_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Average probability of each genetic operator for run x