Nixtla's Open Source Time Series Ecosystem.

Nixtla Community

quick question regarding mlflow. i see the example using statsforecast (<https://nixtlaverse.nixtla.io/statsforecast/docs/tutorials/mlflow.html>). before i try myself: i assume i can just switch statsforecast with mlforecast basically, right?

Sadly no, someone built a mlflow flavor specifically for statsforecast (in the `mlflavors` package). The integration requires several things to be implemented which we don't currently have for ml or neural

wait, so i cant use mlflow togeher with mlforecast at all??

i have not worked with mlflow so much yet and the question is: can i just register and save a model as some kind of general artefact, or is there some specific connection needed between mlflow and the library? because all i know is always library specific like: mlflow.xgboost.log_model, mlflow.sklearn.log_model etc

You can use it for logging but in order to save and load the artifacts for inference we need to define some interfaces which don't currently exist and that's what the xgboost and sklearn modules have