Cumulocity IoT Edge AutoML Regression Error

, , , ,
1

Dear colleagues,

I want to develop a ML model with the AutoML functionality of the ML workbench application to predict house prices. I uploaded the dataset, selected the features and chose the regression problem type. But the training process is failing with this error message.

regressionError
regressionError1353×425 44.2 KB

I have already used the workbench to train a classification problem model, which is working without any errors. So this error is only occurring when choosing a regression problem.

I am using this dataset (California Housing Prices | Kaggle) and Cumulocity IoT Edge version 10.9.0.

Do you have any solutions for me? Maybe through adjusting some advanced training parameters for regression problems?

Best regards,
Julius

2

Hello Julius,

Can you please select either “r2” or “neg_mean_absolute_error” or “neg_mean_squared_error” instead of “Accuracy” as your Scoring metric and retrain the model?

“Accuracy” is the metric for Classification models.

Best,
Vinay

3

Hello Vinay,

That makes sense. My fault. Training is working now. Thank you!

Best regards,
Julius

4

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.