Benchmark experiment warehouse

Share:

Description

warehouse is the constructor function for a benchmark experiment warehouse.

Usage

1
2
warehouse(datasets, B, algorithms = NULL, performances = NULL,
  characteristics = NULL, tests = NULL)

Arguments

datasets

Names of the datasets

B

Number of benchmark runs

algorithms

Names of the candidate algorithms

performances

Names of the performance measures

characteristics

Names of the dataset characteristics

tests

Names of the monitored test measures

Details

A benchmark experiment warehouse collects all data during the execution of a benchmark experiment; see benchmark. Different views (based on the collected data) provide cleaned parts of the data for further analyses.

Implemented views:

  1. viewAlgorithmPerformance(): returns a data frame (S3 class AlgorithmPerformance) with columns samples, datasets, algorithms, performances (factors with the corresponding levels) and the column value with the corresponding performance value.

  2. viewDatasetCharacterization(): returns a data frame (S3 class DatasetCharacterization) with columns samples, datasets, characteristics, value.

  3. viewDatasetBasisCharacterization(): returns a data frame (S3 class DatasetBasisCharacterization) with columns datasets, characteristics, value.

  4. viewTestResult(): returns a data frame (S3 class TestResult) with columns samples, datasets, tests, value.

Value

Proto object with different views (see Details).

See Also

benchmark, as.warehouse