Completion Scoring

../_images/completion-scoring.png

Scores the quality of matrix completion using root mean squared error (RSME) metric.

Signals

Inputs:

  • Fitted fusion graph

    Fitted collective latent data model.

  • Relation

    Relationships between two groups of objects.

Outputs:

  • (None)

Description

This widget assesses the quality of matrix completion based on root mean squared error metric (RMSE). Each row contains scores representing matrix completion quality of different relations. Results for prediction models are in columns.

../_images/CompletionScoring-stamped.png
  1. The RMSE value chart for the input relation matrix.

Example

Completion Scoring widget assesses the quality of matrix completion using the RMSE metric. Connect it with Matrix Sampler to score prediction models (previously learnt on in-sample data) on out-of-the-sample data. You can also use Mean Fuser to get a mean score for latent values.

../_images/CompletionScoring-Example.png