Model-based threshold selection for agricultural linkages / by Christian Arsenault. : CS11-522/2024-1-21E-PDF

"Optimal threshold selection is a critical challenge in probabilistic linkage, with significant implications for the accuracy and reliability of linked datasets. This paper analyzes the performance of the neighbour model, a recently proposed error model which models linkage errors by the number of links from each record. Three threshold selection algorithms utilizing the neighbour model were assessed, highlighting the strengths and limitations of each. Their performance was assessed through simulation studies, which demonstrated that methods using the neighbour model achieved lower relative bias compared to two established methods for threshold selection. Additionally, the practical utility was validated through goodness-of-fit tests conducted on four agricultural datasets, showing the potential of the model for use in the real-world applications"--Abstract, page [2].

Lien permanent pour cette publication :
publications.gc.ca/pub?id=9.966673&sl=1

• Format MARC XML
• Format MARC HTML

Demander des formats alternatifs