Machine Learning API's

[Ambarish_Chitnis (Ambarish)]

I was looking at the ML API's and had a few questions:

1. How do I execute ML jobs on a specific schedule, for example once daily?

2. Is the flow to execute a ML job using the API's as follows: create a job, open the job, add the data feed, start the data feed? An example would really help.

3. How do I specify seasonality (meaning whether the data feed has seasonal data or not and whether it is weekly or monthly)?

4. Which specific unsupervised ML models are used under the hood?

[richcollier]

@Ambarish_Chitnis

• How do I execute ML jobs on a specific schedule, for example once daily?

If the ML job is running in "real-time" (on going in the background) - then the settings of the job's bucket_span and the datafeed's setting of frequency will define the "schedule".

@Ambarish_Chitnis

• Is the flow to execute a ML job using the API's as follows: create a job, open the job, add the data feed, start the data feed? An example would really help.

Example: ml_job_api_example.sh · GitHub

@Ambarish_Chitnis

• How do I specify seasonality (meaning whether the data feed has seasonal data or not and whether it is weekly or monthly)?

ML will automatically determine if and what kind of seasonality appears in your data

@Ambarish_Chitnis

• Which specific unsupervised ML models are used under the hood?

It's a variety of techniques that you can learn about in these videos:

30 Min Meetup w/ Steve Dodson:

Machine Learning in the Elastic Stack

2017 ElasticON:

Elasticsearch Machine Learning and Statistical Methods for Time Series Analysis

2018 ElasticON:

The Math Behind Elastic Machine Learning