Oil spill prediction model using Artificial Intelligence

Abstract

The present investigation focused on the analysis of the variables that led to oil spills that occurred during the years 1900 to 2019 in the state of New York in North America. Under state law and regulations in this country, spills that could contaminate state lands or waters must be reported by the person responsible for the spill (and, in some cases, by anyone with knowledge of the contamination). Each spill record includes: administrative information (region and unique spill code), type of facility, date/time of spill, location, contributing factor, source and cause of spill, type of material spilled, quantity spilled and recovered, bodies of surface water affected, closure date (cleanup activity completed) and Environmental Cost. With this information, Artificial Intelligence prediction models were developed and, due to its greater precision, the Decision Tree model was determined as the most suitable predictor of future ecological disasters.