Classification based on time series data

Dear All,
I have a data set with the following columns:
Date, instance ID, Feature1, 2, n, Class
My task is to train a classifier to predict Class variable which values are Yes, or No. I want to produce new features using the Instance ID as a grouping column in order to exploit the new features which will be the time series of each feature for my classification problem since the time information might make my classifier more accurate. Is there a node that can transform my data into time series according to the latter in order to be used for training a binary classifier?
Many thanks in advance

Hi @npapan -

If I understand you correctly, there’s not a single node that’s going to do what you want - it will probably require a small workflow. That said, it doesn’t sound that complicated. One place to start might be a binning operation using Numeric Binner or Rule Engine nodes.

Still, it’s hard to be too specific without the data. Could you provide an example dataset (sanitized, if it’s business confidential) so folks here could think about the details with you?

1 Like

Dear ScottF many thanks for your reply, I have already solved the problem outside KNIME.


This topic was automatically closed 182 days after the last reply. New replies are no longer allowed.