In my thesis I have shown novel and useful opportunity of utilization of data collected and stored by fleet tracking systems. Handling significant amount of data requires using different types of methods, part of the described algorithms implemented in database environment other parts in frame of other applications. After the preparation for appropriate table structure, most representative time series of road network elements of training data set have been produced by means of clustering processes. Thanks to the classification based on these characteristics, evaluation of average speed in a given hour or missing time series of road network elements of validation set has been done by models. Application of models on normalized data has also been introduced in this document. Accuracy of methods has been verified by various evaluating processes and methods. Description of using representative speed profiles of road network and interfacing with geographic information software terminates my thesis.