[R] date and time format for nonparametric regression

jmoreira@fe.up.pt jmoreira at fe.up.pt
Wed Oct 20 15:04:33 CEST 2004


I have my time series data in Oracle. I want to read the data to test some 
nonparametric regression techniques, such as, neural networks, support 
vectors, random forests, etc. My question is: does someone have experience on 
what is the best way to use the date data? That is, shall I read the date 
(date and time) as a single variable, or is it better to split date and time 
in different variables? The answer is the same for all non parametric 
regression techniques? For time after midnight, the descontinuity from 24 
hours to 0 hours doesn't create problems when using those algorithms?

Thanks in advance

Joao Moreira

More information about the R-help mailing list