For selecting an algorithm from a set of algorithms to solve a serial-batch scheduling problem, we propose in our working paper "Machine Learning based Algorithm Selection and Genetic Algorithms for serial-batch scheduling" to use Machine Learning models for predicting an algorithm based on characteristics of the problem instance.To get an accurate Machine Learning model for this task, a large data set of instances with different "complexities" is needed.This repository contains a data set of serial-batch scheduling problems (with their feature vectors) and several Machine Learning models developed in ["Learning-augmented heuristics for scheduling parallel serial-batch processing machines" (Computers & Operations Research, 2022, 10.1016/j.cor.2022.106122)], and [Uzunoglu, A., Gahm, C. & Tuma, A. A machine learning enhanced multi-start heuristic to efficiently solve a serial-batch scheduling problem. Ann Oper Res (2023). https://doi.org/10.1007/s10479-023-05541-w]. These Machine Learning models are used to perform an efficient grid search in the parameter space of a heuristic. Each Machine Learning-based grid search (MLPP-methods in the first paper and MLRP-methods in the second paper) represents a solution method to solve the serial-batch scheduling problem.The data set comprises two feature vector representations (CF12 with 12 features and AF85 with 85 features) and 17 objective values computed by the solution methods of the aforementioned papers for each problem instance. Please note that the ordering of the features must be maintained when applying the Machine Learning models. The provided serial-batch scheduling problem instances are a subset of the instance set in [Gahm, Christian (2022), “Extended instance sets for the parallel serial-batch scheduling problem with incompatible job families, sequence-dependent setup times, and arbitrary sizes”, Mendeley Data, V2, doi: 10.17632/rxc695hj2k.2] excluding the S (small) instances.

For selecting an algorithm from a set of algorithms to solve a serial-batch scheduling problem, we propose in our working paper "Machine Learning based Algorithm Selection and Genetic Algorithms for serial-batch scheduling" to use Machine Learning models for predicting an algorithm based on characteristics of the problem instance.To get an accurate Machine Learning model for this task, a large data set of instances with different "complexities" is needed.This repository contains a data set of serial-batch scheduling problems (with their feature vectors) and several Machine Learning models developed in ["Learning-augmented heuristics for scheduling parallel serial-batch processing machines" (Computers & Operations Research, 2022, 10.1016/j.cor.2022.106122)], and [Uzunoglu, A., Gahm, C. & Tuma, A. A machine learning enhanced multi-start heuristic to efficiently solve a serial-batch scheduling problem. Ann Oper Res (2023). https://doi.org/10.1007/s10479-023-05541-w]. These Machine Learning models are used to perform an efficient grid search in the parameter space of a heuristic. Each Machine Learning-based grid search (MLPP-methods in the first paper and MLRP-methods in the second paper) represents a solution method to solve the serial-batch scheduling problem.The data set comprises two feature vector representations (CF12 with 12 features and AF85 with 85 features) and 17 objective values computed by the solution methods of the aforementioned papers for each problem instance. Please note that the ordering of the features must be maintained when applying the Machine Learning models. The provided serial-batch scheduling problem instances are a subset of the instance set in [Gahm, Christian (2022), “Extended instance sets for the parallel serial-batch scheduling problem with incompatible job families, sequence-dependent setup times, and arbitrary sizes”, Mendeley Data, V2, doi: 10.17632/rxc695hj2k.2] excluding the S (small) instances.