Constrained Prescriptive Trees via Column Generation

Authors: Shivaram Subramanian, Wei Sun, Youssef Drissi, Markus Ettl4602-4610

AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We demonstrate the efficacy of our method with extensive experiments on both synthetic and real datasets.
Researcher Affiliation Industry IBM Research, Yorktown Heights, NY 10591, USA subshiva, sunw, youssefd, mettl@us.ibm.com
Pseudocode No No pseudocode or algorithm blocks are explicitly presented.
Open Source Code No The paper uses third-party open-source tools (e.g., light GBM) but does not provide a link or explicit statement about the open-sourcing of its own methodology's code.
Open Datasets Yes using a publicly available dataset2 that contains over two years of household level transactions from a group of 2,500 households at a grocery retailer. A processed dataset with 97,295 rows which contain both purchases and no-purchases of strawberries is available online.3
Dataset Splits No The paper mentions dividing data in halves for some experiments and varying training data size, but does not provide specific or consistent train/validation/test split proportions or methodologies for all experiments.
Hardware Specification Yes All experiments were run on an Intel 8-core i7 PC with 32GB RAM.
Software Dependencies Yes CPLEX 20.1 was used to solve the RMP.
Experiment Setup Yes The minimum number of samples per rule for SPMT is set to 10. K = 100 for KSP.