More transparent and explainable machine learning algorithms are required to provide enhanced and sustainable dataset understanding

David A. Wood

Ecological Modelling, 2024, vol. 498, issue C

Abstract: For detailed dataset interrogation and auditing purposes the lack of dataset explainability/transparency of the majority of available machine-learning (ML) models poses limitations. There is a tendency for ML models to focus on prediction speed and accuracy at the expense of transparently revealing dataset relationships. A case is made here to broaden that focus and for ML models to offer alternative configurations tailored to provide more explanations about how individual predictions are derived. Indeed, those striving to achieve sustainable objectives should not rely on opaque ML models and seek transparency as a fundamental objective of good modelling practice (GMP). Doing so tends to boost trust and confidence in the outputs of models relating to complex socio-environmental systems (SES), particularly those being used to potentially justify controversial social, political and ethical decisions. Currently, the transparent open box algorithms (TOB) are the only ML algorithms available that are configured specifically to routinely provide detailed data record relationships for each of their predictions. This study describes the data mining benefits of the Python-coded optimized data-matching TOB algorithms generally, and when applied to environmental datasets characterized by complex non-linear relationships involving many variables.

Keywords: Dataset interrogation; Optimized data matching; Prediction explainability; Forensic dataset interpretability; Transparent open box (TOB) algorithms; Python coded TOB (search for similar items in EconPapers)
Date: 2024
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S0304380024002862
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:ecomod:v:498:y:2024:i:c:s0304380024002862

DOI: 10.1016/j.ecolmodel.2024.110898

Access Statistics for this article

Ecological Modelling is currently edited by Brian D. Fath

More articles in Ecological Modelling from Elsevier
Bibliographic data for series maintained by Catherine Liu ().