简体繁体 English

为什么 ML.net 中需要 Label 列？

[英]Why Label column is needed in ML.net?

原文 2019-11-06 22:40:47 7 3 c#/ machine-learning/ ml.net

I have 20 million records of Dataset to train the model.我有 2000 万条数据集记录来训练 model。

Why Label column I have to add?为什么Label一栏我要加？

3 个解决方案

The "label" is what your model needs to predict. “标签”是您的 model 需要预测的内容。 That is done by training your model with "features" or in other words, information that has relevance to the "label".这是通过使用“特征”或换句话说，与“标签”相关的信息训练您的 model 来完成的。

Label is how ML.NET calls the 'target variable', the one that you want to predict, based on the other variables, called features. Label是 ML.NET 调用“目标变量”的方式，您要根据其他变量进行预测，称为特征。

If you do not give ML.NET a 'label column name' (or a name of your 'target variable' column), it will assume that the label column is called Label .如果您没有为 ML.NET 提供“标签列名称”（或“目标变量”列的名称），它将假定 label 列称为Label 。 Hence the error you are probably seeing Label column 'Label' is not found .因此，您可能看到的错误Label column 'Label' is not found 。

For example lets say we have an image data in the form of pixels in an excel file, if you have a number "5" and it has 50 features.例如，假设我们在 excel 文件中有一个像素形式的图像数据，如果你有一个数字“5”并且它有 50 个特征。 Now 5 is what the " label " holds, and the 50 features are used to learn what a 5 looks like.现在 5 是“ label ”所持有的，50 个特征用于学习 5 的样子。 So label is needed because otherwise it doesn't know what the features belong to.所以需要 label 因为否则它不知道特征属于什么。