Data Mining Car Configurator Clickstream Data to Identify Potential Consumers: A Genetic Algorithm Approach

The Car Configurator (CC) website provided by automotive Original Equipment Manufacturers (OEMs) enables customers to choose from the brand’s portfolio of cars without having to list them all. Afterwards, users move to dealership to formalize the purchase. However, the car they acquired might differ from the one they consulted online. Because there is no record from these deviations, CC data is considered noisy and meaningless. This paper investigates the question of whether valuable information can be extracted from CC clickstream data to aid automotive manufacturers in their operations. The data mining technique of genetic algorithms is employed to identify the characteristics that maximize the correlation between clickstream data and car sales. The findings reveal that the genetic algorithm outperforms the benchmark correlation value and that most frequently occurring elements from sales and webpage data may not be the most effective indicators of potential consumers. The proposed methodology can help identify future clients and target marketing efforts.