Skip to content

Latest commit

 

History

History
237 lines (121 loc) · 10.1 KB

Experiments.md

File metadata and controls

237 lines (121 loc) · 10.1 KB
Raw

Experiments

Datasets

Chocardiogram

Dataset Instances NumAtt ClassProportion
Chocardiogram 131 11 2 Labels (67.2%, 32.8%)
Chocardiogram (clean) 112 11 2 Labels (78.6%, 21.4%)

Noise Visualizacion

Two first principal components visualizacion. Different colors indicate different classes. Crosses indicate those instances with a negative discrimination (estimated 3PL model parameter). Noise Visualization

Original DS

![Ability, Avg Prob of Success and Accuracy per Method](abilityParam/Results (Orig&Clean DS)/Chocarciogram/chocardiogram_2_plotAll.pdf)

![Ability and Avg Prob of Success per Accuracy](abilityParam/Results (Orig&Clean DS)/Chocarciogram/chocardiogram_2_plotAllAcc.pdf)

Clean DS

Dateset without noisy instances (instances with negative discrimination parameter)

![Ability, Avg Prob of Success and Accuracy per Method](abilityParam/Results (Orig&Clean DS)/Chocarciogram/chocardiogram_2_clean_plotAll.pdf)

![Ability and Avg Prob of Success per Accuracy](abilityParam/Results (Orig&Clean DS)/Chocarciogram/chocardiogram_2_clean_plotAllAcc.pdf)

Statlog-heart

Dataset Instances NumAtt ClassProportion
Statlog-heart 270 14 2 Labels (44.4%, 55.6%)
Statlog-heart (clean) 229 14 2 Labels (42.4%, 57.6%)

Noise Visualizacion

Two first principal components visualizacion. Different colors indicate different classes. Crosses indicate those instances with a negative discrimination (estimated 3PL model parameter). Noise Visualization

Original DS

![Ability, Avg Prob of Success and Accuracy per Method](abilityParam/Results (Orig&Clean DS)/Heart-statlog/heart-statlog_plotAll.pdf)

![Ability and Avg Prob of Success per Accuracy](abilityParam/Results (Orig&Clean DS)/Heart-statlog/heart-statlog_plotAllAcc.pdf)

Clean DS

Dateset without noisy instances (instances with negative discrimination parameter)

![Ability, Avg Prob of Success and Accuracy per Method](abilityParam/Results (Orig&Clean DS)/Heart-statlog/heart-statlog_clean_plotAll.pdf)

![Ability and Avg Prob of Success per Accuracy](abilityParam/Results (Orig&Clean DS)/Heart-statlog/heart-statlog_clean_plotAllAcc.pdf)

Hepatitis

Dataset Instances NumAtt ClassProportion
Hepatitis 155 20 2 Labels (79.4%,20.6%)
Hepatitis (Clean) 124 20 2 Labels (90.3%,9.7%)

Noise Visualizacion

Two first principal components visualizacion. Different colors indicate different classes. Crosses indicate those instances with a negative discrimination (estimated 3PL model parameter). Noise Visualization

Original DS

![Ability, Avg Prob of Success and Accuracy per Method](abilityParam/Results (Orig&Clean DS)/hepatitis/06_hepatitis_2_plotAll.pdf)

![Ability and Avg Prob of Success per Accuracy](abilityParam/Results (Orig&Clean DS)/hepatitis/06_hepatitis_2_plotAllAcc.pdf)

Clean DS

Dateset without noisy instances (instances with negative discrimination parameter)

![Ability, Avg Prob of Success and Accuracy per Method](abilityParam/Results (Orig&Clean DS)/hepatitis/06_hepatitis_2_clean_plotAll.pdf)

![Ability and Avg Prob of Success per Accuracy](abilityParam/Results (Orig&Clean DS)/hepatitis/06_hepatitis_2_clean_plotAllAcc.pdf)

Ionosphere

Dataset Instances NumAtt ClassProportion
Ionosphere 323 34 2 Labels (65.3%, 34.7%)
Ionosphere (clean) 351 34 2 Labels (64.1%, 35.9%)

Noise Visualizacion

Two first principal components visualizacion. Different colors indicate different classes. Crosses indicate those instances with a negative discrimination (estimated 3PL model parameter). Noise Visualization

Original DS

![Ability, Avg Prob of Success and Accuracy per Method](abilityParam/Results (Orig&Clean DS)/Ionosphere/ionosphere_plotAll.pdf)

![Ability and Avg Prob of Success per Accuracy](abilityParam/Results (Orig&Clean DS)/Ionosphere/ionosphere_plotAllAcc.pdf)

Clean DS

Dateset without noisy instances (instances with negative discrimination parameter)

![Ability, Avg Prob of Success and Accuracy per Method](abilityParam/Results (Orig&Clean DS)/Ionosphere/ionosphere_clean_plotAll.pdf)

![Ability and Avg Prob of Success per Accuracy](abilityParam/Results (Orig&Clean DS)/Ionosphere/ionosphere_clean_plotAllAcc.pdf)

Parkinsons

Dataset Instances NumAtt ClassProportion
Parkinsons 195 23 2 Labels (75.4%, 24.6%)
Parkinsons (clean) 173 23 2 Labels (82.7%, 17.3%)

Noise Visualizacion

Two first principal components visualizacion. Different colors indicate different classes. Crosses indicate those instances with a negative discrimination (estimated 3PL model parameter). Noise Visualization

Original DS

![Ability, Avg Prob of Success and Accuracy per Method](abilityParam/Results (Orig&Clean DS)/Parkinsons/08_parkinsons_2_plotAll.pdf)

![Ability and Avg Prob of Success per Accuracy](abilityParam/Results (Orig&Clean DS)/Parkinsons/08_parkinsons_2_plotAllAcc.pdf)

Clean DS

Dateset without noisy instances (instances with negative discrimination parameter)

![Ability, Avg Prob of Success and Accuracy per Method](abilityParam/Results (Orig&Clean DS)/Parkinsons/08_parkinsons_2_clean_plotAll.pdf)

![Ability and Avg Prob of Success per Accuracy](abilityParam/Results (Orig&Clean DS)/Parkinsons/08_parkinsons_2_clean_plotAllAcc.pdf)

Seeds

Dataset Instances NumAtt ClassProportion
Seeds 210 8 3 Labels (33.3%, 33.3%, 33.3%)
Seeds (clean) 197 8 3 Labels (32%, 35%, 33%)

Noise Visualizacion

Two first principal components visualizacion. Different colors indicate different classes. Crosses indicate those instances with a negative discrimination (estimated 3PL model parameter). Noise Visualization

Original DS

![Ability, Avg Prob of Success and Accuracy per Method](abilityParam/Results (Orig&Clean DS)/Seeds/20_seeds_3_plotAll.pdf)

![Ability and Avg Prob of Success per Accuracy](abilityParam/Results (Orig&Clean DS)/Seeds/20_seeds_3_plotAllAcc.pdf)

Clean DS

Dateset without noisy instances (instances with negative discrimination parameter)![Ability, Avg Prob of Success and Accuracy per Method](abilityParam/Results (Orig&Clean DS)/Seeds/20_seeds_3_clean_plotAll.pdf)

![Ability and Avg Prob of Success per Accuracy](abilityParam/Results (Orig&Clean DS)/Seeds/20_seeds_3_clean_plotAllAcc.pdf)

Teaching

Dataset Instances NumAtt ClassProportion
Teaching 151 6 3 Labels (35.5%, 35.5%, 29%)
Teaching (clean) 107 6 3 Labels (34.4%, 33.1%, 32.5%)

Noise Visualizacion

Two first principal components visualizacion. Different colors indicate different classes. Crosses indicate those instances with a negative discrimination (estimated 3PL model parameter). Noise Visualization

Original DS

![Ability, Avg Prob of Success and Accuracy per Method](abilityParam/Results (Orig&Clean DS)/Teaching/21_teaching_3_plotAll.pdf)

![Ability and Avg Prob of Success per Accuracy](abilityParam/Results (Orig&Clean DS)/Teaching/21_teaching_3_plotAllAcc.pdf)

Clean DS

Dateset without noisy instances (instances with negative discrimination parameter)

![Ability, Avg Prob of Success and Accuracy per Method](abilityParam/Results (Orig&Clean DS)/Teaching/21_teaching_3_clean_plotAll.pdf)

![Ability and Avg Prob of Success per Accuracy](abilityParam/Results (Orig&Clean DS)/Teaching/21_teaching_3_clean_plotAllAcc.pdf)

Vertebral-column

Dataset Instances NumAtt ClassProportion
Vertebral-column 310 7 3 Labels (16.6%, 53.1%, 30.3%)
Vertebral-column (clean) 271 7 3 Labels (19.4%, 48.4%, 32.3%)

Noise Visualizacion

Two first principal components visualizacion. Different colors indicate different classes. Crosses indicate those instances with a negative discrimination (estimated 3PL model parameter). Noise Visualization

Original DS

![Ability, Avg Prob of Success and Accuracy per Method](abilityParam/Results (Orig&Clean DS)/Vertebral-column/22_vertebral-column_3_plotAll.pdf)

![Ability and Avg Prob of Success per Accuracy](abilityParam/Results (Orig&Clean DS)/Vertebral-column/22_vertebral-column_3_plotAllAcc.pdf)

Clean DS

Dateset without noisy instances (instances with negative discrimination parameter)

![Ability, Avg Prob of Success and Accuracy per Method](abilityParam/Results (Orig&Clean DS)/Vertebral-column/22_vertebral-column_3_clean_plotAll.pdf)

![Ability and Avg Prob of Success per Accuracy](abilityParam/Results (Orig&Clean DS)/Vertebral-column/22_vertebral-column_3_clean_plotAllAcc.pdf)