X-pack Machine Learning


(chaouki) #1

i can use x-pack for neural neutwork ??
why i can’t use x-pack for q-learning ??
should i use elasticsearsh for hadoop MLlib spark ?

(Robert Cowart) #2

The Machine Learning feature of X-Pack is a pre-packaged engine designed to detect anomalies in time-series data. You create jobs, which are run by the engine, and where you have control over which data is analysed (index and fields) and how it is evaluated (the functions - sum, mean, median, etc. - to use). Elastic's goal is not to create a generic framework for ML developers. Instead they want to focus on engines that can quickly provide value and don't require a developer or data scientist to configure.

This doesn't mean that you can't use other techniques against data in stored in Elasticsearch... you simply must develop it yourself using which ever libraries or frameworks you prefer. Data can queried from Elasticsearch by your solution, using either the scroll API or the search_after parameter. Similarly you can store the output of your analysis in an Elasticsearch index, which allows you to leverage Kibana to visualise the results.

(system) #3

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.