tpcp: Tiny Pipelines for Complex Problems - A set of framework independent helpers for algorithms development and evaluation

Küderle A, Richer R, Simpetru R, Eskofier B (2023)


Publication Type: Journal article, Original article

Publication year: 2023

Journal

Book Volume: 8

Pages Range: 4953

Issue: 82

Journal Issue: 82

URI: https://joss.theoj.org/papers/10.21105/joss.04953

DOI: 10.21105/joss.04953

Open Access Link: https://joss.theoj.org/papers/10.21105/joss.04953

Abstract

During algorithm development and analysis researchers regularly use software libraries developed for their specific domain. With such libraries, complex analysis tasks can often be reduced to a couple of lines of code. This not only reduces the amount of implementation required but also prevents errors.

The best developer experience is usually achieved when the entire analysis can be represented with the tools provided by a single library. For example, when an entire machine learning pipeline is represented by a scikit-learn pipeline (Pedregosa et al., 2018), it is extremely easy to switch out and train algorithms. Furthermore, train/test leaks and other methodological errors at various stages in the analysis are automatically prevented – even if the user might not be aware of these issues.

However, if the performed analysis gets too complex, too specific to an application domain, or requires the use of tooling and algorithms from multiple frameworks, developers lose a lot of the benefits provided by individual libraries. In turn, the required skill level and the chance of methodological errors rise.

With tpcp we attempt to overcome the issue by providing higher-level tooling and structure for algorithm development and evaluation that is independent of the frameworks required for the algorithm implementation.

Authors with CRIS profile

Additional Organisation(s)

Related research project(s)

How to cite

APA:

Küderle, A., Richer, R., Simpetru, R., & Eskofier, B. (2023). tpcp: Tiny Pipelines for Complex Problems - A set of framework independent helpers for algorithms development and evaluation. Journal of Open Source Software, 8(82), 4953. https://dx.doi.org/10.21105/joss.04953

MLA:

Küderle, Arne, et al. "tpcp: Tiny Pipelines for Complex Problems - A set of framework independent helpers for algorithms development and evaluation." Journal of Open Source Software 8.82 (2023): 4953.

BibTeX: Download