•
Darts 0.47.0 (~2026-09-06): official MLflow integration to track/compare/persist forecasting experiments; see MLflow quickstart notebook. Not bundled with Prophet/Chronos/TimesFM/sktime posts. Not investment advice.
Darts 0.47.0 (PyPI, public release notes around ~6 Sep 2026; wheel upload on PyPI 4 Sep 2026) adds an official MLflow integration for tracking, comparing, and persisting forecasting-model experiments. See the MLflow quickstart notebook and the unit8co/darts repo. After-work note for Python / data / ML / quant notebooks — not investment advice.
The only course CTA is Introduction to Python Programming.
Before 0.47, many after-work stacks logged Darts runs by hand (model.save + mlflow.log_artifact). 0.47.0 documents a first-party path to track / compare / persist forecasting experiments inside MLflow — same mental model as other sklearn-style experiment trackers, without inventing a private wrapper tonight.
| Item | Detail |
|---|---|
| Version | darts 0.47.0 |
| Headline | Official MLflow integration (track / compare / persist) |
| How-to | 29-MLflow-examples |
| Install | pip install "darts==0.47.0" (+ your usual MLflow env) |
python -m venv .venv-darts047 && source .venv-darts047/bin/activate
pip install "darts==0.47.0" "mlflow>=2"
python -c "import darts, mlflow; print(darts.__version__, mlflow.__version__)"
Then open the official MLflow examples notebook and run one logging path end-to-end (fit → log → compare runs in the MLflow UI). Stop there — do not stand up a shared tracking server on the public internet tonight.
This Hub post is only Darts 0.47 + MLflow. It does not bundle older Hub notes on Prophet maintenance mode, Chronos / TimesFM foundation models, or sktime releases — those stay separate posts.
Pin darts==0.47.0, open the MLflow notebook, log one toy forecast. If Python packaging is the blocker, Introduction to Python Programming.