I am doing some analysis on time series. The time series would consist of 3 channels and contain 5 minute interval data.

What I want is to be able to give it a 1 hour block of 5 minute interval data and it will categorise it based on the entire one hour and picking up some patterns how the time series looks for each of the categories as per the training data.

I have many 1 hour series of 5 minute interval data which is classified to a particular category, and I want to be able to have a deep learning model which can detect the pattern between these samples and be able to determine for new samples which categories they belong to.

****Could you please recommend a type of deep learning model which is capable of this?****

Maybe I don't understand LSTM's but to my understanding they provide a prediction for each point in a time series based on the points that occur before it and would therefore give a series of predictions, where as I want 1 prediction for each hour.

I appreciate any help that you can provide to help me understand this better.



1 Answer 1


Correct me if I am wrong, but the problem you describe here sounds like a classification problem, not a times series forecasting. You, just want to know to what class each 1 hour of data belongs to. If this is the case, you can try using a CNN with 1 dimensional convolutions and 3 channels.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.