You are here:
Enhancing Industrial Process Resilience through Predicting Toxicities of Chemicals Using Random Forests
Citation:
Harten, P. AND Ray Smith. Enhancing Industrial Process Resilience through Predicting Toxicities of Chemicals Using Random Forests. 2019 Enterprise and Infrastructure Resilience Conference, Cincinnati, Ohio, August 12 - 13, 2019.
Impact/Purpose:
The purpose of this study is to investigate the potential risks of alternative chemicals in industrial processes to reduce impact to human health and the environment.
Description:
Industrial enterprises and their chemical processes should be resilient to the changing environments in which they operate. Chemical use may present a risk to their resiliency as changing regulations or societal expectations point to the need to use alternative chemicals. Understanding the potential human health and environmental impacts of chemicals and controlling these risks can help to ensure the resilience of the enterprise and process. The toxicities of all of the chemicals involved should be known to better assess the overall risks. Unfortunately, the toxicities of all chemicals involved in a process may not be known. To overcome this barrier, the unknown toxicities of the chemicals may be predicted from a database of chemicals using many different regression techniques based on similar behavior and molecular structure. The technique to predict the toxicities of chemicals explored in this study is based on building and making predictions on a collection of decision trees and is known as RandomForest. By using this prediction method, the resilience of industrial enterprises and their chemical processes can be better understood.