Maintaining Data Resolution with in-equal sampling frequencies

ZKuiper

Hello, I was given a data set in which the label which I want to predict is sampled between 1-6 hours but the inputs which effect it are sampled every hour. My first thought would be to average the inputs to match the sampling frequency of the output but I was wondering if there was any way to retain all the data incase something like variation rather than just the average value effects the output. I have attached a scrubbed version of the data set if you would like to take a look. Thanks!

Telcontar120

You could retain all the data but you'll need to pivot it so you still end up with only one row per prediction instance (e.g., every 6 hours).  Basically you would be creating extra attributes for the additional sample points.  Once you have done that you could look at doing additional feature generation to capture things like min/max in the sample window, range, std deviation, or other measures of dispersion in the additional sample points.  Take a look at some of the function options in Generate Aggregation for additional ideas.