Exploratory Landscape Analysis Feature Values for the 24 Noiseless BBOB Functions

The dataset contains feature values for the so-called "cheap" features available in the R package flacco [https://github.com/kerschke/flacco] on the 24 noiseless BBOB functions of the COCO benchmark environment [https://github.com/numbbo/coco].

Precisely, the data set covers raw data and median feature values for the following

• Feature Sets: dispersion, \(y\)-distribution, meta model, information content, nearest better clustering and  principal component analysis.
• Functions: 24 noiseless BBOB functions of the COCO benchmark environment [https://github.com/numbbo/coco].
• Instances: First five instances of each function.
• Dimension \(d\):  6 different dimensions, \(d \in \{5,10,15,20,25,30\}\).
• Sample size ("budget") \(b\) for Feature Extraction: 7 different options per dimension, \(b \in \{ 30d, 50d, 100d, 250d, 650d, 800d, 1000d\}\).
• Sampling Distribution: Features are computed using Sobol' low-discrepancy sequence as sampling strategy.
• Repetitions: 100 independent repetitions are available.

Caused by the heavy computation times,  some data records are missing:

• The dispersion feature values are not available for \((d=25,b=1000d), (d=30,b=800d) \) and \((d=30,b=1000d)\);
• Instances 2 to 5 are not available for:
  • \(1000 d\) budgets with \(d \geq 10\)
  • \(800d\) budgets with \(d \geq 15\)
  • \(650d\) budgets with \(d \geq 20\)
  • \(250d\) budgets with \(d = 30\)