Nixtla's Open Source Time Series Ecosystem.

Nixtla Community

Hi <@U05S9S9AQ68>! Yes, you can use the `save`  and `load` functionality to simply load a previously trained model. Every time you have more data you can call the `fit` method on the updated data to continue the training.

Two considerations:
1. It wont be trained only on the new data, but rather sample random timestamps on the whole history. You can drop the oldest history to focus the model on the latest data.
2. You probably dont need to update the model every time you have a new single data point, it will be  expensive/time consuming and the model wont change a lot.