词汇的功能是空的。请检查最小n-gram文档频率 [英] Features for The vocabulary is empty. Please check the Minimum n-gram document frequency

问题描述

35：错误0035：词汇表的功能为空。请检查最小n-gram文档频率。必填但未提供。，错误代码：ModuleExecutionError，Http状态
代码：400，时间戳：星期四，2019年2月28日15:23:51 GMT

< span style ="font-size：0.75em">如何解决此问题？

解决方案

很抱歉听到您遇到此错误。如果没有为给定用户或项目提供功能，则会发生异常。

Azure机器学习中出现此错误，您尝试使用推荐模型得分但无法找到特征向量。

分辨率

使用项目功能或用户功能时，Matchbox推荐程序必须满足某些要求。此错误表示您作为输入提供的用户或项目缺少要素向量。您必须确保每个用户或项目的数据中都有一个向量
的要素。

例如，如果您使用用户的年龄，位置或等功能训练了推荐模型收入，但现在想要为在培训期间未见到的新用户创建分数，您必须为新用户提供一些等效的功能集（即
年龄，位置和收入值），以便做出适当的预测对他们来说

如果您没有这些用户的任何功能，请考虑使用功能工程来生成适当的功能。例如，如果您没有单独的用户年龄或收入值，则可能会生成用于一组用户的近似值。

从推荐模式评分时，您可以使用项目或用户功能只有您之前在培训期间使用了项目或用户功能。有关详细信息，请参阅

得分Matchbox推荐者。

有关Matchbox如何获取的一般信息推荐算法有效，以及如何准备项目功能或用户功能的数据集，请参阅

火车火柴盒推荐人。

问候，

宇通

35: Error 0035: Features for The vocabulary is empty. Please check the Minimum n-gram document frequency. required but not provided., Error code: ModuleExecutionError, Http status code: 400, Timestamp: Thu, 28 Feb 2019 15:23:51 GMT

How to resolve this issue ?

解决方案

Hi,

Sorry to hear you are suffering from this error. Exception occurs if no features were provided for a given user or item.

This error in Azure Machine Learning occurs you are trying to use a recommendation model for scoring but a feature vector cannot be found.

Resolution

The Matchbox recommender has certain requirements that must be met when using either item features or user features. This error indicates that a feature vector is missing for a user or item that you provided as input. You must ensure that a vector of features is available in the data for each user or item.

For example, if you trained a recommendation model using features such as the user's age, location, or income, but now want to create scores for new users who were not seen during training, you must provide some equivalent set of features (namely, age, location and income values) for the new users in order to make appropriate predictions for them.

If you do not have any features for these users, consider feature engineering to generate appropriate features. For example, if you do not have individual user age or income values, you might generate approximate values to use for a group of users.

When you are scoring from a recommendation mode, you can use item or user features only if you previously used item or user features during training. For more information, see Score Matchbox Recommender.

For general information about how the Matchbox recommendation algorithm works, and how to prepare a dataset of item features or user features, see Train Matchbox Recommender.

Regards,

Yutong

这篇关于词汇的功能是空的。请检查最小n-gram文档频率的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！