[jira] [Updated] (SOLR-10991) Support removing top N influential observations in the regress Stream Evaluator

["Cassandra Targett (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/SOLR-10991?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Cassandra Targett updated SOLR-10991:
-------------------------------------
    Component/s: streaming expressions

> Support removing top N influential observations in the regress Stream Evaluator
> -------------------------------------------------------------------------------
>
>                 Key: SOLR-10991
>                 URL: https://issues.apache.org/jira/browse/SOLR-10991
>             Project: Solr
>          Issue Type: New Feature
>      Security Level: Public(Default Security Level. Issues are Public) 
>          Components: streaming expressions
>            Reporter: Joel Bernstein
>            Assignee: Joel Bernstein
>             Fix For: 7.0
>
>
> Influential observations are *outliers* that have a large effect on the slope of a regression line. It is very useful to be able to automatically remove influential observations prior to running a simple regression.
> Syntax:
> {code}
> regress(colA, colB, 10)
> {code}
> The function above regresses colA and colB after removing the top 10 influential observations from the data set.
> The approach taken will be to remove each observation one and at a time and re-run the regression on the data set minus the observation. After each run the difference in model fit will be recorded. After completing the regression runs, N observations that had the highest difference of fit will be removed from the data set. The final regression will be run without those observations.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org