In maxent, when I used a random test percentage of 25 with cross validate replication runs of 10, an error message popped up stating that 'random test percentage must be reset to zero as cross validation runs'. Does it mean that these two can not be used together?. In many papers, I have seen them used together.
Hi,
When using a random test percentage, the software holds those data out for validation purpose and that part of data does not play any role in the model fitting process. However, when using cross-validation all the data is divided into e.g 10 parts and every part is used for model fitting at least once which is a testing procedure and there is no need for keeping another extra random test data set.
I hope this contains your answer.
Cheers,
Roozbeh
Hi,
When using a random test percentage, the software holds those data out for validation purpose and that part of data does not play any role in the model fitting process. However, when using cross-validation all the data is divided into e.g 10 parts and every part is used for model fitting at least once which is a testing procedure and there is no need for keeping another extra random test data set.
I hope this contains your answer.
Cheers,
Roozbeh
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics