aceKnowledge-Based Feature Selection Substantially Enhances Data-Driven Wastewater Treatment Modeling
Data-driven modeling in wastewater treatment is increasingly constrained by the reality of small, high-dimensional data, where the abundant monitoring parameters in small-sized data sets obscure fundamental mechanistic understandings. This study proposes a knowledge-driven feature selection framework that integrates mechanistic insights with statistical correlations to identify the most informative predictive features. Using nitrous oxide (N 2 O) emission prediction at a full-scale plant as a case study, we comp…
Expert-guided and LLM-augmented feature selection improved generalizability of data-driven models for nitrous oxide emission prediction at a full-scale wastewater plant, maintaining temporal dynamics under out-of-distribution high-flow conditions where attention-based deep learning failed.
Attention-based deep-learning feature selection failed to capture N2O emission patterns under out-of-distribution high-flow conditions, and LLM-augmented selection showed lower predictive accuracy with mean R2 0.596 compared to 0.712 for attention-based.
Results are from a single full-scale plant case study for N2O emissions; generalizability to other plants and pollutants not measured in this text
Evidence
- Peer-reviewedEnvironmental Science & Technology2026-07-12
How should this claim be treated?
Truvace Impact Record TRV-2026-0120, v1: “Knowledge-Based Feature Selection Substantially Enhances Data-Driven Wastewater Treatment Modeling.” Truvace, 2026-07-13. /record/TRV-2026-0120 (accessed at citation time). sha256 cd02b5ce51a4cc29…
Calibration history
Every change to this record since certification, in the open. None yet — the reading has held since it entered the record.
Certified into the record
How to verify without trusting this page
Fetch the canonical text of any version from /api/record/TRV-2026-0120 and hash it yourself — for example shasum -a 256 on the saved canonical field. The result must equal content_hash, and each version’s text ends with prev:followed by the prior version’s hash (version 1 chains to 64 zeros). If a single character of any version had been altered since certification, the chain would not reproduce.