I experiment on building electricity power consumption datasets and try to see relationships of the power consumption with weather data and dummy variables that represent time-of-week.

The only thing I do at the moment is OLS regression with Python statsmodels but I am curious if anyone could recommend an unsupervised learning process that could potentially do something similar, that is if an unsurprised learning process can something similar.

This is just a screenshot below, sorry but I use OLS to see the relationships for example the coefficients are in units of building electricity kW where the intercept is the baseload electrical usage and what impacts weather (dewpoint temp in this case) and hours of the day have on building power consumptions.

Strange question, curious about any thoughts! Any unsupervised learning package to try with some sort of visualization of the data relationships for this type of use case would be greatly appreciated.

enter image description here

  • $\begingroup$ I'm not really sure and I've never used it myself but I think Facebook Prophet might be an option. $\endgroup$
    – Erwan
    Commented Mar 10, 2021 at 0:37


Your Answer

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

Browse other questions tagged or ask your own question.