Baseline Model

---

Model Description: Decision Tree Classifier using 11 features to predict outage duration class.

Features Used:

• Quantitative (6): MONTH, ANOMALY.LEVEL, CUSTOMERS.AFFECTED, POPPCT_URBAN, CUSTOMER_DENSITY, POPULATION_DENSITY
• Nominal (5): CLIMATE.CATEGORY, CAUSE.CATEGORY, SEVERITY_CATEGORY, OUTAGE_SEASON, IS_EXTREME_WEATHER

Preprocessing:

• Numeric features: Mean imputation for missing values
• Categorical features: Most frequent imputation + One-hot encoding

Performance:

              precision  recall  f1-score  support
Long             0.80    0.78      0.79     170
Short            0.71    0.73      0.72     126
accuracy                           0.76     296

Assessment: The baseline model achieves 76% accuracy with reasonable precision and recall for both classes. The model performs slightly better on predicting long outages (precision 0.80) than short outages (precision 0.71), which could be valuable for emergency planning purposes. However, there’s room for improvement in the overall performance and class balance.

---