I was wondering how we can set up retraining time-period for Elastic Machine Learning Job?
Suppose, I want to train my model in 10 days data and want to retrain my model every after every 11 days taking 10 days data, is it something that is possible in Elastic?
Overall, How does Elastic Machine Learning Model retraining works?
Hello, I was wondering if there is any update on this?
I want to set up ML Anomaly pipeline. I use data for 1 month, say between April 10- May 10 for training the model, create the anomaly pipeline. But how does the retraining works?
I'm so sorry for the wait! Unfortunately, we don't currently have the option to re-train data frame analytics models as a workflow inside Elastic Stack. But don't worry, we're always looking for ways to improve! I think this is a fantastic idea! I would highly encourage you to fill in an enhancement request and describe how you would like to see this workflow in practice.
When you create a data frame analytics training job using the Kibana wizard, you can use filters to specify the period of interest for training. However, this would create a new model and not update the old model with the new data, so you may need to experiment with the filter settings to get a representative dataset with a focus on recent data.
Finally, you can use the trained models API to substitute the old model with the new one in your ingest pipeline.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.