Data science and machine learning in wastewater intelligence
summary
Kando deploys custom and off-the-shelf sensors plus a lab-sample database and open data to detect and classify pollution events, using a three-step pipeline: candidate generation, expert labeling, and dataset enrichment. Models include regression to proxy expensive analyzers and real-time classifiers to tag pollutant types, and graph- and time-series–based upstream matching to localize sources. Implementations use scikit-learn and networkx alongside proprietary time-series algorithms and yield actionable most-likely source alerts and sensor-cost reductions.
AI-written summary · about
chapters (7)
- 0:00Wastewater intelligence intro
- 2:53Why monitor wastewater now
- 6:43Machine learning role
- 10:12Smart sensor deployment
- 13:22Pollution event classification
- 14:49Locating upstream sources
- 17:16Key takeaways
related talks
25mData science and machine learning in wastewater intelligence 🌳
26mFrom Data to Decision: Monitoring and Improving Water Networks using AI/ML
19mDigital Earth Pacific DEP cloud native approaches to detect mineral extraction
08mUsing machine learning to infuse data operations
37mBig Data Analytics, Machine Learning, Real Time Processing
17mBuilding Water Security From the Bottom Up by Leveraging Big Data
30mAutomated river plastic monitoring Using..
20mAutomating Data Science for the IOT