Today, business intelligence's starting point is the computer-driven applications' by-product, the measured data. The system's inputs, outputs and occasionally the insides are measured and logged throughout data collection.
The observed business processes can be greatly improved by analysing the log files with mathetmatical visualization and the business interpretation of observed phenomenons in the application's context. The topic of the BSc thesis is analysing the required visualization technologies and techniques for this, highlighting the connection of business and mathematical visualizations.
Understanding logs aren't trivial, due to their complexity, thus a data scientist is needed, to analyse the data. The data scientist tries to find correlation, show trends, and make predicitions from the analysed data. To understand a complex business process, an industrial domain expert is also required along with the data scientist, who interprets mathematical results in the business process' context.
A gap appears between the data scientists' and domain experts' visualization language, which creates an obstacle between their cooperation.
This problem often occurs in techical areas, e.g. a low level measured data in a computer system is the resources allocation level, while the business context is the produced service level.
The thesis provides a solution to bridge the gap between data scientists and domain experts, where both sides can practice the interactive analysis in their own visualization language, while the visualizations remain coordinated with the help of the framework I developed.
The thesis first shows the data representation's role and historical evolution. After that, the thesis summarizes the important principles of general data visualization in the modern era. It also presents today's data visualization software categories with heavily-used software examples. It introduces the widely used R data analysing language's abilites and the application-oriented visualization dominant Processing language. In the last chapter a demonstration is shown that the framework is capable of supporting the reproducible data analysis as it is integrated into a new visualization techinque, where the visual analysis steps can be saved and restored.