I know that block bootstrapping is a technique used to resample time series data as it can preserve time dependencies within the data. In particular, if the block size is one month, then data within each month is not reshuffled but the blocks themselves can be reshuffled.
However, I am working with tidymodels and I understand that this has not yet been implemented, so I would have to do this manually. On the other hand, I came across a resampling technique known as group k-fold cross-validation here which can be implemented directly in tidymodels.
The problem is I am not sure whether this is applicable to time series data. To be precise, I would like to know whether group k-fold cross-validation is equivalent to block bootstrapping when it comes to preserving serial correlations. For example, if I group by month, then does that mean that the data within each month is not touched?
Any elaborations on group k-fold cross-validation as well as comparisons to block bootstrapping for the purposes of resampling time series data would be greatly appreciated!
