Invited Session: Smart Technologies
Using Machine Learning Algorithms to Predict Landfill Gas Potential (0.5 CEU)
Tuesday, September 26, 2017
14:00 – 14:30
Moderator: Bernd Leowald, Regional Manager, Stadtreinigung Hamburg A.Ö.R., Germany
Making accurate predictions from complex data can be an overwhelming task, but with widely available data and machine learning providing new and powerful analysis methods, accurate predictions are increasingly easy to make. Existing applications of machine learning in waste management include waste stream sorting, predicting future waste streams, predicting demographic shifts, and predicting equipment maintenance requirements.
Using machine learning algorithms and cloud-computing power, it is possible to build solutions for new and emerging problems in the waste management industry. For example, Amazon.com sales data can be used to predict the waste materials that will come to a facility in the near future or birth records may serve to predict future waste quantities. Using domain knowledge and machine learning, it is possible to reduce both operating costs (such as fuel) and data collection costs. This presentation will explore machine learning algorithms that can be used to find the most important features in making a prediction, potentially reducing the amount of data required to reach an accurate conclusion while also saving time and money. Furthermore, an example of using machine learning to predict how changes in landfill operations might affect long-term gas production, recovery rates and utilization potential will be presented.
By attending this presentation, participants will better be able to discuss how modern computing and data analysis techniques as they apply to waste management, specifically as it applies to reducing costs and improving landfill gas production and utilization potential. The presenter will also introduce machine learning and present example models for discussion.