flacco: Feature-Based Landscape Analysis of Continuous and Constrained Optimization Problems
- If you like our package, please star it on Github.
Introduction
flacco is a collection of features for Explorative Landscape Analysis (ELA) of single-objective, continuous (Black-Box-)Optimization Problems. It allows the user to quantify characteristics of an (unknown) optimization problem's landscape.
Features, which used to be spread over different packages and platforms (R, Matlab, python, etc.), are now combined within this single package. Amongst others, this package contains feature sets, such as ELA, Information Content, Dispersion, (General) Cell Mapping or Barrier Trees.
Furthermore, the package provides a unified interface for all features -- using a so-called feature object and (if required) control arguments. In total, the current release (1.7) consists of 17 different feature sets, which sum up to approximately 300 features.
In addition to the features themselves, this package also provides visualizations, e.g. of the cell mappings, barrier trees or information content:
- 3D-Barrier Tree:
- Cell Mapping:
- Information Content Plot:
Quickstart
If you want to get started quickly, have a look at the flacco tutorial.
library(flacco)
## (1) Create some example-data
X = createInitialSample(n.obs = 500, dim = 2)
f = function(x) sum(sin(x) * x^2 + (x - 0.5)^3)
y = apply(X, 1, f)
## (2) Compute the feature object
feat.object = createFeatureObject(X = X, y = y)
## (3) Have a look at feat.object
print(feat.object)
## (4) Check, which feature sets are available
listAvailableFeatureSets()
## (5) Calculate a specific feature set, e.g. the ELA meta model
featureSet = calculateFeatureSet(feat.object, set = "ela_meta")
Installation Instructions
-
If you want to use the version from CRAN, you should download it from here and ideally install it along with all its dependencies:
install.packages("flacco", dependencies = TRUE)
-
However, if you want to use the development version, you can download it with the following command:
devtools::install_github("kerschke/flacco")
flacco-GUI
We have started to implement a graphical user interface (GUI) for our package. Its online-version is hosted here.
Alternatively, you can start it from within R:
library(flacco)
runFlaccoGUI()
News
- July 15, 2017: Presented flacco (including the GUI) twice at GECCO 2017 @ Berlin:
- "flaccogui: Exploratory Landscape Analysis for Everyone" at the EvoSoft Workshop AND
- within the Advanced Tutorial on "Exploratory Landscape Analysis".
- June 14, 2017: flacco 1.7 has been submitted to CRAN.
- June 09, 2017: started using pkgdown and depsy.
- May 11, 2017: flacco 1.6 has been submitted to CRAN.
- April 28, 2017: flacco 1.5 (including its GUI) has been submitted to CRAN.
- February 27, 2017: flacco 1.4 has been submitted to CRAN.
- July 29, 2016: Presented "The R-Package FLACCO for Exploratory Landscape Analysis with Applications to Multi-Objective Optimization Problems" at the IEEE WCCI / CEC in Vancouver, Canada.
- April 15, 2016: flacco 1.3 has been submitted to CRAN.
- January 22, 2016: flacco 1.2 has been submitted to CRAN.
- November 26, 2015: flacco 1.1 has been submitted to CRAN.
- October 28, 2015: It is done! flacco 1.0 is released on CRAN!
- September 2015: Fixing the last minor issues before submitting version 1.0 to CRAN.
Contact
If you have any suggestions or ideas (e.g. for new features), or if you encounter any problems while running the code, please use the issue tracker or send me an e-mail (kerschke@uni-muenster.de).