| aggregations {mlr} | R Documentation |
Aggregation methods.
Description
- test.mean
Mean of performance values on test sets.
- test.sd
Standard deviation of performance values on test sets.
- test.median
Median of performance values on test sets.
- test.min
Minimum of performance values on test sets.
- test.max
Maximum of performance values on test sets.
- test.sum
Sum of performance values on test sets.
- train.mean
Mean of performance values on training sets.
- train.sd
Standard deviation of performance values on training sets.
- train.median
Median of performance values on training sets.
- train.min
Minimum of performance values on training sets.
- train.max
Maximum of performance values on training sets.
- train.sum
Sum of performance values on training sets.
- b632
Aggregation for B632 bootstrap.
- b632plus
Aggregation for B632+ bootstrap.
- testgroup.mean
Performance values on test sets are grouped according to resampling method. The mean for every group is calculated, then the mean of those means. Mainly used for repeated CV.
- testgroup.sd
Similar to testgroup.mean - after the mean for every group is calculated, the standard deviation of those means is obtained. Mainly used for repeated CV.
- test.join
Performance measure on joined test sets. This is especially useful for small sample sizes where unbalanced group sizes have a significant impact on the aggregation, especially for cross-validation test.join might make sense now. For the repeated CV, the performance is calculated on each repetition and then aggregated with the arithmetic mean.