How to perform machine learning on data that has multiple rows for one response value

machine_learning
predictive_model
data_wrangling

#1

I have streaming alarm data that comes at different time intervals, each alarm has several columns which describes it .Multiple alarms together represent one Root Cause .We want to use Machine learning to predict this root Cause.
I cannot convert my alarm signature into wide format to get one row per root cause as the number or alarms for root cause varies. How can i perform ML??


#2

You will have to convert it into one row. You have to make it a flag variable in case some alarms dont work for specific response. I do not see any other easy approach than creating a summary table at response level(like a pivot) and building a model for the summary table