Related Experiment Video
Updated: Nov 25, 2025
08:27
Author Spotlight: Efficient Image Recognition Using Directional Gradient Histogram Technique and Support Vector Machines
Published on: January 5, 2024
1.4K
Identification the source of fecal contamination for geographically unassociated samples with a statistical
Qiaowen Tan1, Weiying Li1, Xiao Chen2
1State Key Laboratory of Pollution Control and Resource Reuse, Tongji University, Shanghai 200092, China; College of Environmental Science and Engineering, Tongji University, Shanghai 200092, China.
Journal of Hazardous Materials
|December 19, 2020
Summary
Support Vector Machine (SVM) accurately tracks fecal contamination sources in water using bacterial markers. This method is efficient for local and foreign samples, aiding in identifying pollution origins.
Area of Science:
- Environmental microbiology
- Bioinformatics
- Water quality assessment
Background:
- High-throughput sequencing reveals bacterial diversity crucial for fecal contamination source tracking.
- Machine learning models can predict fecal sources, but their performance varies with sample origin.
Purpose of the Study:
- To evaluate the performance of Support Vector Machine (SVM), Random Forest (RF), and Adaboost algorithms for fecal source tracking.
- To identify discriminatory bacterial markers for distinguishing fecal contamination sources.
- To assess the models' accuracy on local and geographically unassociated samples.
Main Methods:
- Utilized high-throughput sequencing to analyze bacterial diversity in fecal samples.
- Applied extremely randomized trees (ExtraTrees) to select discriminatory sequences from Clostridiale, Bacteroidales, and Lactobacillales.
- Trained and compared SVM, RF, and Adaboost classification models for fecal source prediction.
Main Results:
- Discriminatory bacterial markers, comprising 1.51-12.64% of unique sequences, explained 70% of microbiome discrepancies.
- SVM and RF models achieved high accuracy (96.08% and 98.04%) on local samples, outperforming Adaboost (90.20%).
- SVM demonstrated good performance on non-local samples, with minor false positives in closely related groups.
Conclusions:
- SVM is an accurate and time-efficient method for fecal source tracking in contaminated water bodies.
- The SVM model shows potential for analyzing geographically unassociated samples.
- Quantitative PCR (qPCR) remains essential for precise detection of human-specific fecal pollution.
