I have been running the ENMeval package in R for the best possible MaxEnt settings. I came across a tutorial that says we should use the random k-fold data partitioning technique if we have more than 50 occurrence points but for less than 50 occurrence points, the jackknife is the preferred technique. It would be very helpful if someone can explain the reason with reference.
K-fold partitioning is a model replication method where your presence data will be divided into multiple unique calibration and evaluation sets. Final output will be average of all the replications. It is usually better to have many number of points for K-fold partitioning (cross validation). Jackknife helps you understand variable importance.
I don't think there is any reason to avoid using both at the same time. Although, it will increase processing time.
In this study they have used both i.e. cross validation and jackknife for a total of about 500 presence points: Article Projected shifts in the distribution range of Asiatic black ...
and you will find many similar studies
- Badges
- Science topic
- Similar topics
- Filing