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

 

Batch Details
Batch run name smoking_2
Description Smoking 2 on Hilbert
This report file final/smoking_2/smoking_2_report.html
data file /home/cssip/jsherrah/EPrep3/data/smoking_2.dat
Time of completion Thu May 28 00:00:22 1998
Duration of Batch 10 hours, 1 minutes, 49 seconds
Random Seed 706773667 (from clock)
Average generations per run 12.90
Average failed feature creations per run 2650.50
Average fitness evalutions per run 62423.20
Test Set Improvement 31.89 %
 
 

Data Partition
 
Class Training Validation Test Total
0 75 37 39 151
1 359 179 181 719
2 993 497 495 1985
Total 1427 713 715 2855

Summary of Results

  Original Classification Errors (%)
Classifier Training Validation Test
Maximum Likelihood(ML) 56.41 62.55 63.50
Parallelepiped(PPD) 77.37 79.80 88.95
Min. Distance to Means(MDTM) 60.48 59.19 62.80

  EPrep Best-of-run Classification Errors (%)
Run
Training
Validation
Test
McNemar confidence
# Features
# Inputs
# Nodes
Classifier
1 30.34 30.15 74.69 0.000 2 6 108 PPD
2 30.27 30.43 31.05 1.000 5 9 453 PPD
3 29.57 30.15 31.47 1.000 4 9 717 PPD
4 30.13 29.87 30.91 1.000 2 9 393 PPD
5 30.34 30.29 94.55 0.000 2 5 47 PPD
6 30.20 29.87 30.91 1.000 3 4 57 PPD
7 30.41 30.15 31.05 1.000 3 9 658 PPD
8 29.99 30.29 30.91 1.000 2 4 98 PPD
9 30.34 30.29 31.89 1.000 2 8 425 PPD
10 30.20 30.29 30.77 1.000 1 7 222 PPD
Ave. 30.18 (0.23 ) 30.18 (0.18 ) 41.82 (21.86 ) 0.80 (0.40 ) 2.60 (1.11 ) 7.00 (2.00 ) 317.80(235.24) PPD

  Confusion matrix for Best Ever Individual from run 4
Class
Predicted
Total
1
2
3
Ground Truth
1
0 0 39 39
2
0 3 178 181
3
0 4 491 495
Total
0 7 708 715

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

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
smoking_2_gen.m Matlab plot generation function
smoking_2_bogf_ave.{eps,gif} Best-of-generation Fitness, averaged over 10 runs
smoking_2_bogv_ave.{eps,gif} Best-of-generation Validation Set Error, averaged over 10 runs
smoking_2_avef_ave.{eps,gif} Average fitness, averaged over 10 runs
smoking_2_stdf_ave.{eps,gif} Standard deviation of fitness, averaged over 10 runs
smoking_2_nftr_ave.{eps,gif} Average number of features per individual, averaged over 10 runs
smoking_2_nnode_ave.{eps,gif} Average number of nodes per individual, averaged over 10 runs
smoking_2_nint_ave.{eps,gif} Average number of introns per individual, averaged over 10 runs
smoking_2_ntrl_ave.{eps,gif} Average number of RAT trials per individual, averaged over 10 runs
smoking_2_optimp_ave.{eps,gif} Average improvement in fitness due to optimisation, averaged over 10 runs
smoking_2_opprob_ave.{eps,gif} Average probability of each genetic operator, averaged over 10 runs
smoking_2_run_x.dat Binary data file containing results of run x (read by Matlab functions)
smoking_2_bor_x.prep 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Best-of-run individual for run x
smoking_2_run_x.corr 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Feature correlation file for run x
smoking_2_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
smoking_2_bogf_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Best-of-generation Fitness for run x
smoking_2_bogv_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Best-of-generation Validation Set Error for run x
smoking_2_avef_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Average fitness for run x
smoking_2_stdf_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Standard deviation of fitness for run x
smoking_2_nftr_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Average number of features per individual for run x
smoking_2_nnode_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Average number of nodes per individual for run x
smoking_2_nint_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Average number of introns per individual for run x
smoking_2_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
smoking_2_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
smoking_2_opprob_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Average probability of each genetic operator for run x