In our rushing world the demand, that services should be adjusted to the people, has been raised. One of its consequences, that they do not have to sit in front of the television waiting for their favourite show, but also they can watch them in an appropriate time. Video on Demand services offer a solution for this situation, where the contents are available in any time of day.
This thesis contains trend analysing and prediction of consumption based on data of a Video on Demand content provider. In making tasks, purchase data of the full-length movies happened in periodical demolition – hourly, daily, weekly, monthly demolition –, then making of prediction of daily consumption with the usage of various environmental variables. For all this, basis of data mining and linear regression applying are presented, furthermore the detailed demonstration of Video on Demand service, as well as the usage of RapidMiner datamining and analyser software, which was proved to be excellent complemented by Excel table manipulator software except for one time. In the case of exception, a data processing program code was written in language C#, which program made the necessary subtask. For the model of prediction, the required data were collected and converted, and the demolition of movies according to genres, then based on these, performance of linear regression model applied on the aggregated daily consumptions is presented. Finally, the summary of results of tasks can be read.