Eureka! Institutional Repository

In recent years, multiple factors have contributed to change the way of life and everyday life of many people in general. The effects of the global financial crisis, the coronavirus, the energy crisis and the war have set in motion remarkable changes both in Greece and in general in the EU. One of the sectors that was decisively affected by the above events is the automotive industry. By studying the parameters that affect car sales, several governments put competent and worthy people in several positions who have the ability to propose innovative ways to support sales of new vehicles that are harmless for the environment, not only in Greece but in all EU countries. At the same time, there is a strong desire from prospective buyers for new generation cars. Some focus on electrics, several on hybrids as well as gasoline engines, fewer now on diesel engines. Due to the strong demand for new cars, the factories are short of important products for the new cars, resulting in significant delays in the delivery of the car. The purpose of this thesis is to create a data set with features of new cars, with the ambition to build a useful and flexible knowledge for prospective buyers. With the help of knowledge mining as well as machine learning, clustering algorithms were applied to find cars with similar characteristics. Then, for the automatic determination of fuel type, classification algorithms are applied with the characteristics of the cars and those that affect the performance of the car. Experiments were carried out that can be applied by experts to draw valuable conclusions. Several machine learning technologies and libraries were leveraged so that we were able to make predictions and conclude to beneficial results.