Summary of 3_Linear
July 10, 2020 ยท View on GitHub
Logistic Regression (Linear)
- explain_level: 2
Validation
- validation_type: split
- train_ratio: 0.75
- shuffle: True
- stratify: True
Optimized metric
logloss
Training time
3.2 seconds
Metric details
| score | threshold | |
|---|---|---|
| logloss | 0.381399 | nan |
| auc | 0.857771 | nan |
| f1 | 0.636203 | 0.309388 |
| accuracy | 0.827396 | 0.40705 |
| precision | 0.915033 | 0.819295 |
| recall | 1 | 0.000610889 |
| mcc | 0.514338 | 0.335185 |
Confusion matrix (at threshold=0.309388)
| Predicted as negative | Predicted as positive | |
|---|---|---|
| Labeled as negative | 4199 | 745 |
| Labeled as positive | 489 | 1079 |
Learning curves
Coefficients
| feature | Learner_1 |
|---|---|
| capital-gain | 2.28219 |
| education-num | 0.843846 |
| age | 0.468591 |
| sex | 0.468299 |
| hours-per-week | 0.369091 |
| capital-loss | 0.279562 |
| race | 0.104163 |
| education | 0.0546121 |
| fnlwgt | 0.0545988 |
| native-country | 0.0173909 |
| occupation | -0.00958272 |
| workclass | -0.102386 |
| relationship | -0.154081 |
| marital-status | -0.358737 |
| intercept | -1.51172 |
Permutation-based Importance
SHAP Importance
SHAP Dependence plots
Dependence (Fold #1)
SHAP Decision plots
Top-10 Worst decisions for class 0 (Fold #1)
Top-10 Best decisions for class 0 (Fold #1)
Top-10 Worst decisions for class 1 (Fold #1)
Top-10 Best decisions for class 1 (Fold #1)
