Nowadays, almost everyone uses some kind of navigation device while driving. Thanks to this, we can easily save the route’s information (distance, time, speed). Furthermore, we can get more information if the running application is connected to the OBD interface. These data include the various parameters of the engine and the fuel consumption.
We can use these recorded data to support the driver with different informations about the path. An ordinary person often has to make the same route (for example while commuting from home to work, and back) over and over. There are several alternative routes (fastest, shortest) for the commuters, but if we consider the fuel consumption we can support the driver with even more possibilities.
In my thesis, I transformed the raw information from the GPS device and OBD into data mining issues, then I solved the problems with the CRISP-DM (Cross Industry Standard Process for Data Mining) methodology. After the preparation and cleaning of the log files, I identified the proper pathes, then for each group I indentified their typical parameters. I made the models and the analysis with the Rapidminer open source predictive analytics platform.
During the evaluation of the data mining process, I tried to find out the answer which is the most prefered route during daily commuting, considering the parameters from the recorded data which affect the fuel consumption. These parameters contains the route’s distance, the average speed and stop times. Finally, I revealed the differences in consumption between parts of the day and also recommended specific routes where we can spare fuel. While I was making this thesis, I have learnt the basics of the intelligent transport systems and had a deeper look into data mining.