Azure ML Studio - Predicting Monthly Sales

Question

Hi there,

I’m trying to use Azure ML Studio to try and predict monthly sales performance. Below is a mock sample of my data uploaded onto Azure. What I want to do is predict the monthly sales for June and July (red font below). The only data available at this stage is the monthly sales from previous months.   

I’m a beginner to ML Studio, and have set up the pipeline as per below screenshot. I’ve chosen the boosted decision tree, and set a split ratio of 70%.

Because I want to predict June, and July, I want the model to look at the preceding column values for each month in deriving its prediction. However, in the train model pop-up, I can only select one column – in the below screenshot, I’ve selected May 2023. Furthermore, when I run this model, I get a Compute job failed, with error in Split Data (see attached Split Data error message Split Data Error Message.txt).

Can someone help me figure out what went wrong? Thank you!

Answer 1

@ning johnny zhao I think the error as per your log:

"ModuleExceptionMessage:InvalidDataset: Dataset is not valid. Input data type EXCEL_WORKBOOK is unrecognized."

indicates that the data module connected to split data isn't a compatible format which split data module can use. Generally, data passed to split data needs to have selected columns with normalization to ensure the right split is used for training and the remaining for testing or validation. Based on the above screen shot it looks like you have not used some of these modules. If you have just started using Azure ML studio, I would recommend going through MS learn courses that have guided steps on using designer to build jobs and endpoints. In this case, a simple regression experiment can be built using this course. The course contains a hands-on lab that further explains the significance of using each step.

Also, check out other learn courses to use Azure ML from the library.

I hope this helps!!

If this answers your query, do click Accept Answer and Yes for was this answer helpful. And, if you have any further query do let us know.