This is the performance analyzer of Iterative Optimization Heuristics (IOHs), e.g., Evolutionary Algorithms. IOHanalyzer takes as input the benchmark data from the user and provides very detailed analysis on, e.g., the empirical running time.
- A web-based Graphical User Interface (GUI).
- Interactive plotting in two perspectives (fixed-budget and fixed-target) of both performance and tracked parameters.
- Statistical evaluation methods.
Rprogramming interfaces in the backend.
It takes as input:
- COCO/BBOB data format (see https://github.com/numbbo/coco),
- IOHprofiler data format, which is motivated and modified from COCO/BBOB data format (Please read the data format section for specifications on the supported data format), or
- Nevergrad data format (explained in https://github.com/facebookresearch/nevergrad).
It is built on:
It is available through:
- a free online service available at http://iohprofiler.liacs.nl, and
- a local installation of the package.
RAs IOHanalyzer is written as a
Renvironment has to be installed first. The binary file and installation manual for R can be found at https://cran.r-project.org/.
orcarequired to download plotly figures. Please see https://github.com/plotly/orca for the installation instruction.
Please start up a
R console and install the stable version as:
which is maintained on CRAN (Comprehensive R Archive Network).
The lastest development is always hosted on Github. In case you’d like to try out this version, the
R package devtool is needed:
Start up the Graphical User Interface
The IOHanalyzer package can be installed and loaded using the following commands:
Have fun! For the complete reference on usage, please continue reading this documentation.