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

 

Batch Details
Batch run name segment_3
Description Segment 3 on Hilbert
This report file final/segment_3/segment_3_report.html
data file /home/cssip/jsherrah/EPrep3/data/segment_3.dat
Time of completion Tue Jun 2 05:46:15 1998
Duration of Batch 20 hours, 33 minutes, 37 seconds
Random Seed 578978304 (from clock)
Average generations per run 33.40
Average failed feature creations per run 1164.90
Average fitness evalutions per run 106907.50
Test Set Improvement 12.80 %
 
 

Data Partition
 
Class Training Validation Test Total
0 165 82 83 330
1 165 82 83 330
2 165 83 82 330
3 165 83 82 330
4 165 82 83 330
5 165 82 83 330
6 165 83 82 330
Total 1155 577 578 2310

Summary of Results

  Original Classification Errors (%)
Classifier Training Validation Test
Maximum Likelihood(ML) 16.19 17.85 19.20

  EPrep Best-of-run Classification Errors (%)
Run
Training
Validation
Test
McNemar confidence
# Features
# Inputs
# Nodes
Classifier
1 4.24 4.16 6.40 1.000 9 9 111 ML
2 6.15 6.41 8.82 1.000 6 10 51 ML
3 7.10 7.63 7.79 1.000 3 7 46 ML
4 5.71 6.07 6.40 1.000 7 7 31 ML
5 5.54 5.03 6.75 1.000 8 8 38 ML
6 5.63 5.37 7.09 1.000 4 10 70 ML
7 6.23 6.24 7.44 1.000 4 10 61 ML
8 5.80 5.37 7.79 1.000 7 7 13 ML
9 5.71 5.89 7.44 1.000 5 5 12 ML
10 4.42 5.03 6.40 1.000 7 11 309 ML
Ave. 5.65 (0.79 ) 5.72 (0.90 ) 7.23 (0.75 ) 1.00 (0.00 ) 6.00 (1.84 ) 8.40 (1.80 ) 74.20 (82.95 ) ML

  Confusion matrix for Best Ever Individual from run 1
Class
Predicted
Total
1
2
3
4
5
6
7
Ground Truth
1
80 0 1 0 2 0 0 83
2
0 82 0 1 0 0 0 83
3
0 0 67 1 14 0 0 82
4
1 0 0 79 2 0 0 82
5
1 0 8 3 71 0 0 83
6
0 0 0 2 0 81 0 83
7
0 0 1 0 0 0 81 82
Total
82 82 77 86 89 81 81 578

Average Operator Probabilities
Operator Average Probability
Delete-Feature Mutation 0.095
Add-Feature Mutation 0.089
Hoist Mutation 0.094
Truncate Mutation 0.104
Swap Mutation 0.102
One-Symbol Mutation 0.097
All-Nodes Mutation 0.099
One-Node Mutation 0.101
Grow Mutation 0.103
High-Level Crossover 0.116
 

Number of Run Terminations attributed to each Criterion
Termination Criterion Number of Terminations
TP Criterion 0
GL Criterion 6
Max. Generations 4
Client Abort 0
Zero Validation Error 0
Total 10
 
 

Related Data Files
Description Filename
segment_3_gen.m Matlab plot generation function
segment_3_bogf_ave.{eps,gif} Best-of-generation Fitness, averaged over 10 runs
segment_3_bogv_ave.{eps,gif} Best-of-generation Validation Set Error, averaged over 10 runs
segment_3_avef_ave.{eps,gif} Average fitness, averaged over 10 runs
segment_3_stdf_ave.{eps,gif} Standard deviation of fitness, averaged over 10 runs
segment_3_nftr_ave.{eps,gif} Average number of features per individual, averaged over 10 runs
segment_3_nnode_ave.{eps,gif} Average number of nodes per individual, averaged over 10 runs
segment_3_nint_ave.{eps,gif} Average number of introns per individual, averaged over 10 runs
segment_3_ntrl_ave.{eps,gif} Average number of RAT trials per individual, averaged over 10 runs
segment_3_optimp_ave.{eps,gif} Average improvement in fitness due to optimisation, averaged over 10 runs
segment_3_opprob_ave.{eps,gif} Average probability of each genetic operator, averaged over 10 runs
segment_3_run_x.dat Binary data file containing results of run x (read by Matlab functions)
segment_3_bor_x.prep 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Best-of-run individual for run x
segment_3_run_x.corr 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Feature correlation file for run x
segment_3_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
segment_3_bogf_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Best-of-generation Fitness for run x
segment_3_bogv_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Best-of-generation Validation Set Error for run x
segment_3_avef_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Average fitness for run x
segment_3_stdf_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Standard deviation of fitness for run x
segment_3_nftr_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Average number of features per individual for run x
segment_3_nnode_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Average number of nodes per individual for run x
segment_3_nint_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Average number of introns per individual for run x
segment_3_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
segment_3_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
segment_3_opprob_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Average probability of each genetic operator for run x