Descriptive and predictive mining on road accidents data
František Babič, Karin Zuskacova · 2016
The paper describes one possibility of how to use the collected data about road accidents to mine frequent patterns and important factors causing different types of accidents. For this purpose was used the real data sample representing road accidents in the United Kingdom (UK) during the years 2005 to 2015. This sample includes more than 1 million records described by 67 attributes divided into three datasets: accidents, casualties and vehicles. Two alternatives were selected as the most suitable: predictive mining through decision trees algorithms and descriptive mining resulted into interesting association rules generated by Apriori algorithm. Obtained results are plausible in comparison with other similar works.