ORCA: A Matlab/Octave Toolbox for Ordinal Regression

Authors: Javier Sánchez-Monedero, Pedro A. Gutiérrez, María Pérez-Ortiz

JMLR 2019 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental ORCA (Ordinal Regression and Classification Algorithms) is a Matlab/Octave framework that implements and integrates different ordinal classification algorithms and specifically designed performance metrics. The framework simplifies the task of experimental comparison to a great extent, allowing the user to: (i) describe experiments by simple configuration files; (ii) automatically run different data partitions; (iii) parallelize the executions; (iv) generate a variety of performance reports and (v) include new algorithms by using its intuitive interface.
Researcher Affiliation Academia Javier S anchez-Monedero EMAIL School of Journalism, Media and Culture CardiffUniversity CardiffCF10 1FS, Wales, United Kingdom Pedro A. Guti errez EMAIL Department of Computer Science and Numerical Analysis University of C ordoba C ordoba, 14071, Spain Mar ıa P erez-Ortiz EMAIL Department of Computer Science University College London London WC1E 6EA, United Kingdom
Pseudocode No The paper includes 'Sample Code' in Section 4, which provides Matlab/Octave code snippets demonstrating API usage. While formatted like code, it is presented as executable example code rather than a generalized algorithm block or pseudocode explicitly detailing a method's steps.
Open Source Code Yes Source code, binaries, documentation, descriptions and links to data sets and tutorials (including examples of educational purpose) are available at https://github.com/ayrna/orca.
Open Datasets Yes Source code, binaries, documentation, descriptions and links to data sets and tutorials (including examples of educational purpose) are available at https://github.com/ayrna/orca. We also provide several small data sets for code testing purposes and a list of 44 ordinal data sets of different characteristics that can be used for algorithm comparison.
Dataset Splits Yes It processes every experiment by fitting a model with the training data (including model selection through cross-validation), evaluating the test error and creating performance reports of all the performance metrics... For example, if we perform 30 repetitions of a hold-out design for 10 data sets, we need at least 300 calls to the fit and predict functions, which can be run in parallel.
Hardware Specification No The paper mentions 'Matlab and Octave parallelisation toolboxes' and 'HTCondor distributed computing environment' for running experiments, but does not provide specific hardware details such as CPU/GPU models, memory, or specific cloud instance types.
Software Dependencies No The paper states 'ORCA (Ordinal Regression and Classification Algorithms) is a Matlab/Octave framework' and mentions using 'Matlab and Octave parallelisation toolboxes' and a 'Matlab wrapper for LIBLINEAR'. However, no specific version numbers are provided for Matlab, Octave, their toolboxes, or LIBLINEAR.
Experiment Setup No The paper mentions 'a specific hyper-parameter configuration of the selected classifier' and that the framework records 'hyper-parameter values'. It also refers to 'model selection through cross-validation'. However, no concrete hyperparameter values (e.g., learning rate, batch size, specific optimizer settings, number of epochs) are explicitly provided in the main text.