Comparative Strategies for Handling Imbalanced QSAR Regression: A Case Study on hERG Activity Prediction
Abstract
Data imbalance frequently occurs in QSAR modeling, yet most studies focus on classification tasks, and few address regression tasks, as it is challenging to determine which regions of the data are sparse or dense. Data-level and algorithm-level imbalance-handling strategies can have different effects on model performance. In this study, we systematically compare various strategies across GCN and fingerprint-based machine learning models. To evaluate model performance more meaningfully, we partitioned the target region into several subregions based on both domain knowledge and data distribution. We focused on the region of greatest practical interest. Our results demonstrate that selecting models based on this interest-region metric leads to more reliable performance than using the total region metric. We also observed that different imbalance-handling strategies can have divergent effects depending on the model type. These findings highlight the importance of region-aware evaluation in regression QSAR modeling and provide practical guidance for selecting appropriate strategies for imbalanced datasets.
Related
- An automated ranking platform for machine learning regression models for meat spoilage prediction using multi-spectral imaging and metabolic profiling
- Finlay–Wilkinson random regression for yield and yield stability prediction in cereals
- Sensitivity of machine learning regression models to data structure and quality in crop yield prediction
- Comparison of data mining algorithms and regression method for prediction of live body weight using biometric traits of non-descript indigenous chickens.
- The Effects of Intermittent Hypoxic Training Strategies on Maximal Oxygen Uptake in Healthy Humans: A Meta-analysis with Meta-regression
BibTeX
@misc{sung2026comparative,
title = {{Comparative Strategies for Handling Imbalanced QSAR Regression: A Case Study on hERG Activity Prediction}},
author = {Ming-Yen Sung and Hsin Liu and Tung-Sheng Cho and Cheng-Wei Lee and Yufeng J. Tseng},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.01709},
eprint = {2611.01709},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.01709},
}
RIS
TY - UNPB
TI - Comparative Strategies for Handling Imbalanced QSAR Regression: A Case Study on hERG Activity Prediction
AU - Ming-Yen Sung
AU - Hsin Liu
AU - Tung-Sheng Cho
AU - Cheng-Wei Lee
AU - Yufeng J. Tseng
PY - 2026
DA - 2026/11/01
PB - arXiv
DO - 10.48550/arxiv.2611.01709
UR - https://arxiv.org/abs/2611.01709
AB - Data imbalance frequently occurs in QSAR modeling, yet most studies focus on classification tasks, and few address regression tasks, as it is challenging to determine which regions of the data are sparse or dense. Data-level and algorithm-level imbalance-handling strategies can have different effects on model performance. In this study, we systematically compare various strategies across GCN and fingerprint-based machine learning models. To evaluate model performance more meaningfully, we partitioned the target region into several subregions based on both domain knowledge and data distribution. We focused on the region of greatest practical interest. Our results demonstrate that selecting models based on this interest-region metric leads to more reliable performance than using the total region metric. We also observed that different imbalance-handling strategies can have divergent effects depending on the model type. These findings highlight the importance of region-aware evaluation in regression QSAR modeling and provide practical guidance for selecting appropriate strategies for imbalanced datasets.
ER -