MarkLogic 12 EA 1 Product Documentationcts:linear-model
cts:linear-model( $values as cts:reference*, [$options as xs:string*], [$query as cts:query?], [$forest-ids as xs:unsignedLong*] ) as math:linear-model?
Summary
Returns a linear model that fits the frequency-weighted data set.
The input data is a co-occurrence, formed from the specified value
lexicons. The length of the input lexicon sequence should be 2, as
currently only simple linear regression model is supported. The
first lexicon should be the value of the dependent variable while
the other lexicon should be the value of the independent variable.
This function works like math:linear-model except each
pair in the input lexicons is counted cts:frequency times.
This function performs the calculation in parallel in all data nodes
then aggregates the values.
| Parameters | |
|---|---|
| values | References to two range indexes. The types of the range indexes must be numeric. If the size of this sequence is not 2, the function returns the empty sequence. |
| options |
Same as the "options" parameter in cts:aggregate.
|
| query |
Same as the "query" parameter in cts:aggregate.
|
| forest-ids |
Same as the "forest-ids" parameter in cts:aggregate.
|
Example
xquery version "1.0-ml";
(:
This query assumes range indexes of type unsignedLong
with positions are configured in the database for both
xval and yval. It generates some sample data and then
performs the aggregation in a separate transaction.
:)
for $i in (1 to 10)
let $j := 2*$i + 1
let $uri := concat($i, ".xml")
return
xdmp:document-insert($uri, <a><xval>{$i}</xval><yval>{$j}</yval></a>);
cts:linear-model(
(cts:element-reference(xs:QName("yval"), ("type=unsignedLong")),
cts:element-reference(xs:QName("xval"), ("type=unsignedLong"))),
"item-frequency")
=>
math:linear-model(
<math:linear-model intercept="1" coefficients="2" rsquared="1"
xmlns:xs="http://www.w3.org/2001/XMLSchema"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xmlns:math="http://marklogic.com/xdmp/math"/>)