Pegasus Bank Risk Models
Retrained on sklearn 1.7.2 from the translator111 questionnaire training script.
Models
risk_model.pkl — RandomForestClassifier (100 trees, max_depth=10)
- Input: 8 features [age, experience, income_bucket, goals, time_horizon, risk_tolerance, risk_behavior, allocation%]
- Output: int class {0: Conservative, 1: Moderate, 2: Growth, 3: Aggressive}
- Train accuracy: 0.9760
- n_features_in_: 8
allocation_model.pkl — GradientBoostingRegressor for portfolio allocation %
Trained with seed=42 on 2000 synthetic samples (deterministic, reproducible).
Reproduced from D:/translator111/translator/train_questionnaire_models.py.
Usage on server: place both .pkl files at /opt/ai_agent/data/models/.
