InfinityLoop208_Finance_Aiml_F11

[InfinityLoop208]

Feature Implemented
Model Versioning System — automatically saves trained models with semantic version tags (e.g. v1_base, v2_tuned) and maintains a JSON metadata registry tracking hyperparameters, training date, and accuracy metrics per version.

Changes Made

src/models/model_utils.py — fully implemented with the following functions:

save_model() — saves .pkl file with version tag + writes metadata
load_model() — loads a specific version by tag
list_versions() — prints a formatted table of all saved versions
rollback() — rolls back to any previously saved version

Output Description

Models are saved to models_saved/model_<version_tag>.pkl
A models_saved/metadata.json file is auto-generated and updated on every save, example:

json{
"v1_base": {
"filename": "model_v1_base.pkl",
"training_date": "2026-04-09 14:32:10",
"hyperparameters": {"n_estimators": 100, "random_state": 42},
"metrics": {"accuracy": 0.91, "precision": 0.88, "recall": 0.85}
}
}