评分模型Score Model

本文介绍 Azure 机器学习设计器(预览版)中的一个模块。This article describes a module in Azure Machine Learning designer (preview).

使用此模块可以通过经训练的分类或回归模型生成预测值。Use this module to generate predictions using a trained classification or regression model.

如何使用How to use

  1. 将“评分模型”模块添加到管道。Add the Score Model module to your pipeline.

  2. 连接经训练的模型和包含新输入数据的数据集。Attach a trained model and a dataset containing new input data.

    数据应采用与所使用的经训练模型类型兼容的格式。The data should be in a format compatible with the type of trained model you are using. 输入数据集的架构通常还应与用于训练模型的数据的架构相匹配。The schema of the input dataset should also generally match the schema of the data used to train the model.

  3. 提交管道。Submit the pipeline.

结果Results

使用评分模型生成一组分数后:After you have generated a set of scores using Score Model:

  • 若要生成一组用于评估模型准确性(性能)的指标,可将评分数据集连接到评估模型模块。To generate a set of metrics used for evaluating the model's accuracy (performance), you can connect the scored dataset to Evaluate Model,
  • 右键单击模块并选择“可视化”可查看结果的示例。Right-click the module and select Visualize to see a sample of the results.

分数(或预测值)可以采用多种不同的格式,具体取决于模型和输入数据:The score, or predicted value, can be in many different formats, depending on the model and your input data:

  • 对于分类模型,分数模型输出类的预测值,以及预测值的概率。For classification models, Score Model outputs a predicted value for the class, as well as the probability of the predicted value.
  • 对于回归模型,评分模型仅生成预测数值。For regression models, Score Model generates just the predicted numeric value.

将评分发布为 Web 服务Publish scores as a web service

评分的一个常见用途是在预测 Web 服务中返回输出。A common use of scoring is to return the output as part of a predictive web service. 有关详细信息,请参阅此教程,了解如何在 Azure 机器学习设计器中基于管道部署实时终结点。For more information, see this tutorial on how to deploy a real-time endpoint based on a pipeline in Azure Machine Learning designer.

后续步骤Next steps

请参阅 Azure 机器学习的可用模块集See the set of modules available to Azure Machine Learning.