Szczegóły publikacji
Opis bibliograficzny
Version 3.0.0 – AccuClass: a tool for confusion-matrix-based metrics in machine learning, remote sensing, and spatial exposure analysis / P. KRAMARCZYK, B. HEJMANOWSKA // SoftwareX [Dokument elektroniczny]. — Czasopismo elektroniczne ; ISSN 2352-7110 . — 2026 — vol. 35 art. no. 102871, s. 1-6. — Wymagania systemowe: Adobe Reader. — Bibliogr. s. 6, Abstr. — Publikacja dostępna online od: 2026-07-15
Autorzy (2)
Słowa kluczowe
Dane bibliometryczne
| ID BaDAP | 169353 |
|---|---|
| Data dodania do BaDAP | 2026-09-16 |
| Tekst źródłowy | URL |
| DOI | 10.1016/j.softx.2026.102871 |
| Rok publikacji | 2026 |
| Typ publikacji | artykuł w czasopiśmie |
| Otwarty dostęp |  |
| Czasopismo/seria | SoftwareX |
Abstract
This software update extends the existing AccuClass tool by introducing a Streamlit-based graphical user interface (GUI), simplifying installation through a precompiled wheel compatible with pipx, and synchronizing documentation across the README and built-in help system. The update preserves full backward compatibility while enabling non-technical users to compute and explore a wide range of confusion-matrix-based metrics without command-line interaction. Beyond software improvements, the paper demonstrates a non-standard application of AccuClass in the context of flood exposure analysis. A spatial cross-tabulation between land-use categories and a binary flood extent is used to derive exposure and spatial association metrics that are formally equivalent to classical accuracy measures but intentionally reinterpreted outside a classification framework. Metrics such as Producer's Accuracy, User's Accuracy, F1 score, and Matthews correlation coefficient are shown to provide meaningful indicators of land-use exposure, flood composition, and spatial association, while global measures dominated by true negatives are reported only for completeness. The example highlights the flexibility of confusion-matrix-based metrics and illustrates how AccuClass can support spatial exposure analysis and decision-oriented studies beyond traditional accuracy assessments in remote sensing and machine learning.
