AutoGRD is a meta-learning method for algorithm recommendation. Here is a python3 implementation of the paper "AutoGRD: Model Recommendation Through Graphical Dataset Representation" by Noy Cohen-Shapira, Lior Rokach, Bracha Shapira, Gilad Katz and Roman Vainshtein. The paper published at CIKM 2019 conference and won the best research paper award. https://dl.acm.org/citation.cfm?id=3357384.3357896
The order of the code:
- RandomForestclassification
- Co-occurence_computation
- Orca_algorithm
- Embeeding_creation
Our work is based on two external sources:
- Implementation of the Orca algorithm, was taken from: http://www.biolab.si/supp/orca/.
- Implementation of the Graphlet Correlation Distance, was taken from: http://www0.cs.ucl.ac.uk/staff/natasa/GCD/.
Requirements: spark 2.4.4 python3 sklearn