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

 

Batch Details
Batch run name australian_1
Description Australian Credit 1 on Craigieburn
This report file results/final/australian_1/australian_1_report.html
data file /home/users/jsherrah/EPrep3/data/australian_1.dat
Time of completion Thu May 14 02:15:58 1998
Duration of Batch 13 hours, 53 minutes, 11 seconds
Random Seed 539107465 (from clock)
Average generations per run 13.40
Average failed feature creations per run 5500.50
Average fitness evalutions per run 63521.10
Test Set Improvement 27.17 %
 
 

Data Partition
 
Class Training Validation Test Total
0 191 96 96 383
1 154 76 77 307
Total 345 172 173 690

Summary of Results

  Original Classification Errors (%)
Classifier Training Validation Test
Parallelepiped(PPD) 37.39 41.28 39.88
Min. Distance to Means(MDTM) 34.49 30.23 37.57

  EPrep Best-of-run Classification Errors (%)
Run
Training
Validation
Test
McNemar confidence
# Features
# Inputs
# Nodes
Classifier
1 17.97 10.47 10.40 1.000 2 3 13 MDTM
2 17.39 11.63 11.56 1.000 1 1 5 MDTM
3 17.39 11.63 11.56 1.000 6 8 54 MDTM
4 17.39 12.79 10.98 1.000 1 2 6 MDTM
5 17.39 11.63 11.56 1.000 1 1 11 MDTM
6 17.39 11.63 11.56 1.000 2 3 5 MDTM
7 18.55 11.63 11.56 1.000 1 3 7 MDTM
8 17.39 11.63 11.56 1.000 1 4 17 MDTM
9 17.39 11.63 11.56 1.000 4 6 50 MDTM
10 17.39 11.63 11.56 1.000 1 1 1 MDTM
Ave. 17.57 (0.37 ) 11.63 (0.52 ) 11.39 (0.37 ) 1.00 (0.00 ) 2.00 (1.61 ) 3.20 (2.18 ) 16.90 (18.10 ) MDTM

  Confusion matrix for Best Ever Individual from run 1
Class
Predicted
Total
1
2
Ground Truth
1
80 16 96
2
2 75 77
Total
82 91 173

Average Operator Probabilities
Operator Average Probability
Delete-Feature Mutation 0.099
Add-Feature Mutation 0.101
Hoist Mutation 0.099
Truncate Mutation 0.100
Swap Mutation 0.099
One-Symbol Mutation 0.100
All-Nodes Mutation 0.099
One-Node Mutation 0.101
Grow Mutation 0.102
High-Level Crossover 0.101
 

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

Related Data Files
Description Filename
australian_1_gen.m Matlab plot generation function
australian_1_bogf_ave.{eps,gif} Best-of-generation Fitness, averaged over 10 runs
australian_1_bogv_ave.{eps,gif} Best-of-generation Validation Set Error, averaged over 10 runs
australian_1_avef_ave.{eps,gif} Average fitness, averaged over 10 runs
australian_1_stdf_ave.{eps,gif} Standard deviation of fitness, averaged over 10 runs
australian_1_nftr_ave.{eps,gif} Average number of features per individual, averaged over 10 runs
australian_1_nnode_ave.{eps,gif} Average number of nodes per individual, averaged over 10 runs
australian_1_nint_ave.{eps,gif} Average number of introns per individual, averaged over 10 runs
australian_1_ntrl_ave.{eps,gif} Average number of RAT trials per individual, averaged over 10 runs
australian_1_optimp_ave.{eps,gif} Average improvement in fitness due to optimisation, averaged over 10 runs
australian_1_opprob_ave.{eps,gif} Average probability of each genetic operator, averaged over 10 runs
australian_1_run_x.dat Binary data file containing results of run x (read by Matlab functions)
australian_1_bor_x.prep 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Best-of-run individual for run x
australian_1_run_x.corr 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Feature correlation file for run x
australian_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
australian_1_bogf_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Best-of-generation Fitness for run x
australian_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
australian_1_avef_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Average fitness for run x
australian_1_stdf_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Standard deviation of fitness for run x
australian_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
australian_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
australian_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
australian_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
australian_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
australian_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