I am currently making a weather forecast prediction using time series and have that temperature depends on the time of the year and can make models using this, however it also has dependence on longitude and latitude and these variables are not time dependent, how can I include these variables in a model to forecast future temperatures.

my data set

I also have multiple temp recordings for each date and don't know how to use these.

  • $\begingroup$ A question must have a question mark, please edit your post. $\endgroup$
    – Eduard
    Feb 20, 2023 at 19:07

1 Answer 1


It depends on what exactly you are trying to do. Since the temperature for one location is a univariate time series you could fit an ARIMA model for each one and discard the geo-coordinates (given the time series at each point are long enough). However, with the geo-coordinates at hand you could e.g. build features that include the distance between points in the model. Close-by points will have similar changes in temperature and including a measure of proximity might improve predictions. I can imagine here to use modelling techniques from Panel data analyis where one has repeated measurements of the same entities/locations.


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.