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

 

Batch Details
Batch run name diabetes_3
Description Diabetes 3 on Hilbert
This report file final/diabetes_3/diabetes_3_report.html
data file /home/cssip/jsherrah/EPrep3/data/diabetes_3.dat
Time of completion Thu May 28 13:09:05 1998
Duration of Batch 3 hours, 23 minutes, 16 seconds
Random Seed 168339586 (from clock)
Average generations per run 35.20
Average failed feature creations per run 1289.50
Average fitness evalutions per run 132475.30
Test Set Improvement 13.02 %
 
 

Data Partition
 
Class Training Validation Test Total
0 250 125 125 500
1 134 67 67 268
Total 384 192 192 768

Summary of Results

  Original Classification Errors (%)
Classifier Training Validation Test
Parallelepiped(PPD) 61.46 55.73 64.58
Min. Distance to Means(MDTM) 36.46 39.58 37.50

  EPrep Best-of-run Classification Errors (%)
Run
Training
Validation
Test
McNemar confidence
# Features
# Inputs
# Nodes
Classifier
1 25.78 23.96 40.62 0.248 2 4 13 PPD
2 24.48 22.40 25.52 0.999 1 5 14 PPD
3 22.92 22.40 22.40 1.000 3 8 29 MDTM
4 24.48 22.92 23.96 1.000 6 8 191 MDTM
5 25.00 21.88 24.48 0.999 3 5 23 MDTM
6 23.44 21.88 26.56 0.994 4 8 114 MDTM
7 24.74 24.48 30.73 0.948 4 6 32 MDTM
8 24.74 23.44 21.88 1.000 2 6 32 MDTM
9 24.74 22.92 40.10 0.284 8 8 79 PPD
10 25.00 22.40 29.69 0.966 3 3 10 PPD
Ave. 24.53 (0.77 ) 22.86 (0.82 ) 28.59 (6.47 ) 0.84 (0.29 ) 3.60 (1.96 ) 6.10 (1.76 ) 53.70 (55.47 ) MDTM

  Confusion matrix for Best Ever Individual from run 5
Class
Predicted
Total
1
2
Ground Truth
1
99 26 125
2
21 46 67
Total
120 72 192

Average Operator Probabilities
Operator Average Probability
Delete-Feature Mutation 0.095
Add-Feature Mutation 0.108
Hoist Mutation 0.098
Truncate Mutation 0.095
Swap Mutation 0.096
One-Symbol Mutation 0.103
All-Nodes Mutation 0.094
One-Node Mutation 0.096
Grow Mutation 0.095
High-Level Crossover 0.119
 

Number of Run Terminations attributed to each Criterion
Termination Criterion Number of Terminations
TP Criterion 0
GL Criterion 3
Max. Generations 7
Client Abort 0
Zero Validation Error 0
Total 10
 
 

Related Data Files
Description Filename
diabetes_3_gen.m Matlab plot generation function
diabetes_3_bogf_ave.{eps,gif} Best-of-generation Fitness, averaged over 10 runs
diabetes_3_bogv_ave.{eps,gif} Best-of-generation Validation Set Error, averaged over 10 runs
diabetes_3_avef_ave.{eps,gif} Average fitness, averaged over 10 runs
diabetes_3_stdf_ave.{eps,gif} Standard deviation of fitness, averaged over 10 runs
diabetes_3_nftr_ave.{eps,gif} Average number of features per individual, averaged over 10 runs
diabetes_3_nnode_ave.{eps,gif} Average number of nodes per individual, averaged over 10 runs
diabetes_3_nint_ave.{eps,gif} Average number of introns per individual, averaged over 10 runs
diabetes_3_ntrl_ave.{eps,gif} Average number of RAT trials per individual, averaged over 10 runs
diabetes_3_optimp_ave.{eps,gif} Average improvement in fitness due to optimisation, averaged over 10 runs
diabetes_3_opprob_ave.{eps,gif} Average probability of each genetic operator, averaged over 10 runs
diabetes_3_run_x.dat Binary data file containing results of run x (read by Matlab functions)
diabetes_3_bor_x.prep 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Best-of-run individual for run x
diabetes_3_run_x.corr 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Feature correlation file for run x
diabetes_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
diabetes_3_bogf_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Best-of-generation Fitness for run x
diabetes_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
diabetes_3_avef_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Average fitness for run x
diabetes_3_stdf_x.{eps,gif} 
x = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
Standard deviation of fitness for run x
diabetes_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
diabetes_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
diabetes_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
diabetes_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
diabetes_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
diabetes_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