Skip to content

How is the final score of a LambdaMart model calculated? #460

Open
@meg261995

Description

@meg261995

So I am using Elasticsearch LTR plugin for ranking purposes and I have trained an sltr model "my-model" with 4 features.. The below is the explanation of how the score is calculated got by the "explain" parameter of the query. I understood the inner details but having a trouble in understanding how the final score 2.777. Usually the description mentions if the score is either a "product of ","sum of", "min or max". Here I am not able to understand. Can anyone please help me with this:

{ "value": 2.777724,
"description": " LtrModel: LambdaMART using features:",
"details": [ { "value": 12.079079, "description": "Feature 0('feat1')", "details": [ ..... ] },
{ "value": 17.195917, "description": "Feature 1(feat2):", "details": [ ... ] },
{ "value": 0, "description": "Feature 2(feat3):", "details": [ .... ] },
{ "value": 0.8042648, "description": "Feature 3(feat4):", "details": [ .... ] } }

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions