Skip to content

Latest commit

 

History

History
509 lines (320 loc) · 16.2 KB

dados.md

File metadata and controls

509 lines (320 loc) · 16.2 KB

teste do classicador

SMOTE = GradientBoostingClassifier

melhor resultado entre os usados

precision recall f1-score support

0       0.02      0.18      0.04        11
1       0.98      0.87      0.92       675

accuracy                           0.85       686

macro avg 0.50 0.52 0.48 686 weighted avg 0.97 0.85 0.91 686

pre       rec       spe        f1       geo       iba       sup

0       0.02      0.18      0.87      0.04      0.40      0.15        11
1       0.98      0.87      0.18      0.92      0.40      0.17       675

avg / total 0.97 0.85 0.19 0.91 0.40 0.17 686

Acurácia balanceada: 0.5235016835016835 [[ 2 9] [ 91 584]]

RandomOverSampler = GradientBoostingClassifier

precision recall f1-score support

0       0.00      0.00      0.00        11
1       0.98      0.97      0.98       675

accuracy                           0.95       686

macro avg 0.49 0.48 0.49 686 weighted avg 0.97 0.95 0.96 686

pre       rec       spe        f1       geo       iba       sup

0       0.00      0.00      0.97      0.00      0.00      0.00        11
1       0.98      0.97      0.00      0.98      0.00      0.00       675

avg / total 0.97 0.95 0.02 0.96 0.00 0.00 686

Acurácia balanceada: 0.48444444444444446 [[ 0 11] [ 21 654]]

ADASYN = GradientBoostingClassifier

precision recall f1-score support

0       0.01      0.09      0.02        11
1       0.98      0.87      0.92       675

accuracy                           0.86       686

macro avg 0.50 0.48 0.47 686 weighted avg 0.97 0.86 0.91 686

pre       rec       spe        f1       geo       iba       sup

0       0.01      0.09      0.87      0.02      0.28      0.07        11
1       0.98      0.87      0.09      0.92      0.28      0.09       675

avg / total 0.97 0.86 0.10 0.91 0.28 0.09 686

Acurácia balanceada: 0.48175084175084176 [[ 1 10] [ 86 589]]

Melhores modelos de classificação

Perceptron Acurácia balanceada: 0.5453872053872054

GradientBoostingClassifier Acurácia balanceada: 0.5235016835016835

RadiusNeighborsClassifier Acurácia balanceada: 0.5044444444444445

BernoulliNB Acurácia balanceada: 0.5022222222222222

1.1. Linear Models

Perceptron

precision recall f1-score support

0       0.02      0.82      0.04        11
1       0.99      0.27      0.43       675

accuracy                           0.28       686

macro avg 0.50 0.55 0.23 686 weighted avg 0.97 0.28 0.42 686

pre       rec       spe        f1       geo       iba       sup

0       0.02      0.82      0.27      0.04      0.47      0.24        11
1       0.99      0.27      0.82      0.43      0.47      0.21       675

avg / total 0.97 0.28 0.81 0.42 0.47 0.21 686

Acurácia balanceada: 0.5453872053872054 [[ 9 2] [491 184]]

1.3. Kernel ridge regression

não fiz

1.4. Support Vector Machines

make_pipeline(StandardScaler(), SVC(gamma='auto'))

precision recall f1-score support

0       0.01      0.18      0.03        11
1       0.98      0.79      0.87       675

accuracy                           0.78       686

macro avg 0.50 0.48 0.45 686 weighted avg 0.97 0.78 0.86 686

pre       rec       spe        f1       geo       iba       sup

0       0.01      0.18      0.79      0.03      0.38      0.13        11
1       0.98      0.79      0.18      0.87      0.38      0.15       675

avg / total 0.97 0.78 0.19 0.86 0.38 0.15 686

Acurácia balanceada: 0.48350168350168354 [[ 2 9] [145 530]]

svm.SVC

precision    recall  f1-score   support

0       0.01      0.09      0.01        11
1       0.98      0.80      0.88       675

accuracy                           0.79       686

macro avg 0.49 0.45 0.45 686 weighted avg 0.97 0.79 0.87 686

pre       rec       spe        f1       geo       iba       sup

0       0.01      0.09      0.80      0.01      0.27      0.07        11
1       0.98      0.80      0.09      0.88      0.27      0.08       675

avg / total 0.97 0.79 0.10 0.87 0.27 0.08 686

Acurácia balanceada: 0.4476767676767677 [[ 1 10] [132 543]]

NuSVC

precision recall f1-score support

0       0.01      0.09      0.01        11
1       0.98      0.80      0.88       675

accuracy                           0.78       686

macro avg 0.49 0.44 0.45 686 weighted avg 0.97 0.78 0.87 686

pre       rec       spe        f1       geo       iba       sup

0       0.01      0.09      0.80      0.01      0.27      0.07        11
1       0.98      0.80      0.09      0.88      0.27      0.08       675

avg / total 0.97 0.78 0.10 0.87 0.27 0.08 686

Acurácia balanceada: 0.44323232323232326 [[ 1 10] [138 537]]

LinearSVC

precision recall f1-score support

0       0.02      0.18      0.03        11
       1       0.98      0.82      0.89       675

accuracy                           0.81       686

macro avg 0.50 0.50 0.46 686 weighted avg 0.97 0.81 0.88 686

pre       rec       spe        f1       geo       iba       sup

0       0.02      0.18      0.82      0.03      0.39      0.14        11
      1       0.98      0.82      0.18      0.89      0.39      0.16       675

avg / total 0.97 0.81 0.19 0.88 0.39 0.16 686

Acurácia balanceada: 0.4990572390572391 [[ 2 9] [124 551]]

1.5. Stochastic Gradient Descent

SGDClassifier

precision recall f1-score support

0       0.00      0.00      0.00        11
1       0.98      0.93      0.96       675

accuracy                           0.91       686

macro avg 0.49 0.46 0.48 686 weighted avg 0.97 0.91 0.94 686

pre       rec       spe        f1       geo       iba       sup

0       0.00      0.00      0.93      0.00      0.00      0.00        11
1       0.98      0.93      0.00      0.96      0.00      0.00       675

avg / total 0.97 0.91 0.01 0.94 0.00 0.00 686

Acurácia balanceada: 0.46444444444444444 [[ 0 11] [ 48 627]]

StandardScaler

precision recall f1-score support

0       0.00      0.00      0.00        11
1       0.98      0.93      0.96       675

accuracy                           0.91       686

macro avg 0.49 0.46 0.48 686 weighted avg 0.97 0.91 0.94 686

pre       rec       spe        f1       geo       iba       sup

0       0.00      0.00      0.93      0.00      0.00      0.00        11
1       0.98      0.93      0.00      0.96      0.00      0.00       675

avg / total 0.97 0.91 0.01 0.94 0.00 0.00 686

Acurácia balanceada: 0.46444444444444444 [[ 0 11] [ 48 627]]

1.6. Nearest Neighbors

KNeighborsClassifier

precision recall f1-score support

0       0.00      0.00      0.00        11
1       0.98      0.88      0.93       675

accuracy                           0.87       686

macro avg 0.49 0.44 0.47 686 weighted avg 0.97 0.87 0.92 686

pre       rec       spe        f1       geo       iba       sup

0       0.00      0.00      0.88      0.00      0.00      0.00        11
1       0.98      0.88      0.00      0.93      0.00      0.00       675

avg / total 0.97 0.87 0.01 0.92 0.00 0.00 686

Acurácia balanceada: 0.44222222222222224 [[ 0 11] [ 78 597]]

RadiusNeighborsClassifier

precision recall f1-score support

0       0.02      1.00      0.03        11
1       1.00      0.01      0.02       675

accuracy                           0.02       686

macro avg 0.51 0.50 0.02 686 weighted avg 0.98 0.02 0.02 686

pre       rec       spe        f1       geo       iba       sup

0       0.02      1.00      0.01      0.03      0.09      0.01        11
 1       1.00      0.01      1.00      0.02      0.09      0.01       675

avg / total 0.98 0.02 0.98 0.02 0.09 0.01 686

Acurácia balanceada: 0.5044444444444445 [[ 11 0] [669 6]]

NearestCentroid

precision recall f1-score support

0       0.01      0.36      0.03        11
1       0.98      0.60      0.74       675

accuracy                           0.59       686

macro avg 0.50 0.48 0.38 686 weighted avg 0.97 0.59 0.73 686

pre       rec       spe        f1       geo       iba       sup

0       0.01      0.36      0.60      0.03      0.47      0.21        11
1       0.98      0.60      0.36      0.74      0.47      0.22       675

avg / total 0.97 0.59 0.37 0.73 0.47 0.22 686

Acurácia balanceada: 0.4795959595959596 [[ 4 7] [273 402]]

1.7. Gaussian Processes

GaussianProcessRegressor

1.8. Cross decomposition

Nao fiz, pois o modelo PLSCanonical pode não ser a melhor escolha para problemas de classificação, pois ele é mais comumente usado para análise de correlação entre conjuntos de dados multivariados.

1.9. Naive Bayes

GaussianNB

precision recall f1-score support

0       0.01      0.18      0.02        11
1       0.98      0.69      0.81       675

accuracy                           0.68       686

macro avg 0.50 0.44 0.41 686 weighted avg 0.97 0.68 0.80 686

pre       rec       spe        f1       geo       iba       sup

0       0.01      0.18      0.69      0.02      0.35      0.12        11
1       0.98      0.69      0.18      0.81      0.35      0.13       675

avg / total 0.97 0.68 0.19 0.80 0.35 0.13 686

Acurácia balanceada: 0.4368350168350168 [[ 2 9] [208 467]]

BernoulliNB

precision recall f1-score support

0       0.02      1.00      0.03        11
 1       1.00      0.00      0.01       675

accuracy                           0.02       686

macro avg 0.51 0.50 0.02 686 weighted avg 0.98 0.02 0.01 686

pre       rec       spe        f1       geo       iba       sup

0       0.02      1.00      0.00      0.03      0.07      0.00        11
1       1.00      0.00      1.00      0.01      0.07      0.00       675

avg / total 0.98 0.02 0.98 0.01 0.07 0.00 686

Acurácia balanceada: 0.5022222222222222 [[ 11 0] [672 3]]

CategoricalNB

precision recall f1-score support

0       0.00      0.00      0.00        11
       1       0.98      1.00      0.99       675

accuracy                           0.98       686

macro avg 0.49 0.50 0.49 686 weighted avg 0.97 0.98 0.97 686

pre       rec       spe        f1       geo       iba       sup

0       0.00      0.00      1.00      0.00      0.00      0.00        11
      1       0.98      1.00      0.00      0.99      0.00      0.00       675

avg / total 0.97 0.98 0.02 0.97 0.00 0.00 686

Acurácia balanceada: 0.49777777777777776 [[ 0 11] [ 3 672]]

1.10. Decision Trees

1.10.1. Classification

precision recall f1-score support

0       0.01      0.09      0.02        11
1       0.98      0.88      0.93       675

accuracy                           0.87       686

macro avg 0.50 0.48 0.47 686 weighted avg 0.97 0.87 0.91 686

pre       rec       spe        f1       geo       iba       sup

0       0.01      0.09      0.88      0.02      0.28      0.07        11
1       0.98      0.88      0.09      0.93      0.28      0.09       675

avg / total 0.97 0.87 0.10 0.91 0.28 0.09 686

Acurácia balanceada: 0.4847138047138047 [[ 1 10] [ 82 593]]

1.11. Ensembles: Gradient boosting, random forests, bagging, voting, stacking

1.11.1. HistGradientBoostingClassifier

precision recall f1-score support

0       0.00      0.00      0.00        11
1       0.98      0.92      0.95       675

accuracy                           0.90       686

macro avg 0.49 0.46 0.47 686 weighted avg 0.97 0.90 0.93 686

pre       rec       spe        f1       geo       iba       sup

0       0.00      0.00      0.92      0.00      0.00      0.00        11
1       0.98      0.92      0.00      0.95      0.00      0.00       675

avg / total 0.97 0.90 0.01 0.93 0.00 0.00 686

Acurácia balanceada: 0.4585185185185185 [[ 0 11] [ 56 619]]

1.11. RandomForestClassifier

precision recall f1-score support

0       0.00      0.00      0.00        11
1       0.98      0.96      0.97       675

accuracy                           0.94       686

macro avg 0.49 0.48 0.49 686 weighted avg 0.97 0.94 0.96 686

pre       rec       spe        f1       geo       iba       sup

0       0.00      0.00      0.96      0.00      0.00      0.00        11
1       0.98      0.96      0.00      0.97      0.00      0.00       675

avg / total 0.97 0.94 0.02 0.96 0.00 0.00 686

Acurácia balanceada: 0.48 [[ 0 11] [ 27 648]]

1.11.3. Bagging meta-estimator

precision recall f1-score support

0       0.01      0.09      0.02        11
 1       0.98      0.89      0.94       675

accuracy                           0.88       686

macro avg 0.50 0.49 0.48 686 weighted avg 0.97 0.88 0.92 686

pre       rec       spe        f1       geo       iba       sup

0       0.01      0.09      0.89      0.02      0.28      0.07        11
1       0.98      0.89      0.09      0.94      0.28      0.09       675

avg / total 0.97 0.88 0.10 0.92 0.28 0.09 686

Acurácia balanceada: 0.4921212121212121 [[ 1 10] [ 72 603]]

1.16. Probability calibration

1.17. Neural network models (supervised)

1.17.2. Classification

precision recall f1-score support

0       0.01      0.18      0.02        11
1       0.98      0.67      0.80       675

accuracy                           0.66       686

macro avg 0.49 0.43 0.41 686 weighted avg 0.96 0.66 0.79 686

pre       rec       spe        f1       geo       iba       sup

0       0.01      0.18      0.67      0.02      0.35      0.12        11
1       0.98      0.67      0.18      0.80      0.35      0.13       675

avg / total 0.96 0.66 0.19 0.79 0.35 0.13 686

Acurácia balanceada: 0.42720538720538725 [[ 2 9] [221 454]]