aggregations {mlr}

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.

See Also

Aggregation