当我在Excel中运行azure ML实验时，我收到错误

Question

Error! {"error":{"code":"LibraryExecutionError","message":"Module execution encountered an internal library error.","details":[{"code":"TableSchemaColumnCountMismatch","target":" (AFx Library)","message":"data: The table column count (0) must match the schema column count (17)."}]}}

你能帮我解决这个问题吗？

Answer 1

The problem is that the original data file you used to train your Machine Learning Model does not have the same number of columns as the test file that you are using in Excel to make your predictions using the Test Dataset. If you look at the Schema of the Predictive Algorithm you are calling within Excel you will see which columns are "extra" in the original Dataset. Although you can select columns out with the Azure Machine Learning module, "Select Columns in Dataset", this does not prevent the excluded columns from being detected by the Web service you deploy and then ending up in the schema of the Predictive Algorithm you deployed in Azure ML when you call it in Excel. When you are in Excel with the test Dataset that you will use your predictive web service on, make sure the columns in the test Dataset match one to one with the dataset you used to train your machine learning model (with the exception of the scored Label column that will have the results of the predictive algorithm on the Azure Machine Learning website). Edit the training Dataset by removing columns you are going to exclude with the Select Columns in Dataset Module and then upload this edited dataset and use it to retrain your Machine Learning Model before you deploy it. This should get it to work.