0

I came through many blogs still cant understand about cross_validation and performance_metrics

cross_validation(m, initial='2008 days', period='30 days', horizon = '30 days')

I have a Dataset length of 2038 days data

The above code means it will train with 2008 days (initial) data and test with next 30 days (horizon) data in one iteration because period is 30. am i correct ?

first of all i want to know about cross_validation and its parameter and performance_metrics.

Mokey D Luffy
  • 266
  • 2
  • 13
  • 1
    in `cross_validataion` more important is that it will split data into train and test data in different way in every iteration. In normal validation you split data into train and test data only once and always use the same data for train and the same data for test. See images in Wikipedia: [Cross-validation_(statistics)](https://en.wikipedia.org/wiki/Cross-validation_(statistics)) - for example https://en.wikipedia.org/wiki/Cross-validation_(statistics)#/media/File:K-fold_cross_validation_EN.svg – furas Jul 30 '22 at 12:33
  • You can find the description of the logic in this post - [link](https://stackoverflow.com/questions/70070734/understanding-fbprophet-cross-validation/70816207#70816207) – annabitel Sep 16 '22 at 07:05
  • Does this answer your question? [understanding fbprophet cross\_validation](https://stackoverflow.com/questions/70070734/understanding-fbprophet-cross-validation) – annabitel Sep 16 '22 at 07:11

0 Answers0