|
|
Data fusion
Janíková, Veronika ; Hovorka, Tomáš (advisor) ; Anděl, Jiří (referee)
This bachelor's thesis deals with a data fusion, which is a one of the possible solutions to the common problem of data availability in praxis. In the first part, practical use of data fusion, especially in marketing, fundamental algorithms and data fusion problems are discussed. The main part of this thesis deals with the so-called "unconstrained statistical fusion". Firstly, one of the possible processes of this type of fusion is described theoretically in detail. This process involves branching into four different types of data fusion. Next, a method of theoretical evaluating the quality of the general fusion model is designed using statistical indicators. The practical part of the thesis contents processes of four types of unconstrained statistical fusion and their evaluation which both are programmed in statistical program R. Furthermore, the fusion is applied to our artificially generated database and also to a real data collected in praxis by Czech public opinion research company Median. In the very last part of the thesis, the results of fusions applied to these two databases are interpreted, evaluated and discussed.
|
guest ::
