Browsing by Subject "customer data"

Time-dependent hierarchical data is a complex type of data that is difficult to visualize in a clear manner. It can be found in many real-life situations, for example in customer analysis, but the best practices for visualizing this type of data are not commonly known in business world. This thesis focuses on visualizing changes over time in hierarchical customer data using the Plotly Python Graphing Library and is written as an assignment for a Finnish company. The thesis consists of a literature survey and experimental part. The literature survey introduces the most common hierarchical visualization methods, and the different possible encoding techniques for adding time dimension on top of these hierarchical visualization methods. Moreover, the pros and cons of different visualization techniques and encodings are discussed about. In the experimental part of the thesis, visualization prototypes are designed using the Plotly Python Graphing Library. A company customer data set of the commissioning company is partitioned into hierarchical customer segments by a hierarchical industrial classification TOL 2008, and changes over time in a continuous variable are visualized by these segments. Two hierarchical visualization techniques: the sunburst chart and treemap, are used to create two prototype versions, and the combination of color, typography, and interaction is used to encode time dimension in these prototypes. The same prototypes are also exploited to visualize customer segments by an artificial hierarchy created by combining multiple categorical features into a hierarchical structure. The prototypes are validated in the commissioning company by arranging an end user study and expert review. Concerning the prototypes by the industrial classification: According to the end users and experts, both prototype versions are very useful and well-implemented. Among the end users, there was no significant difference in which one of these prototype versions is faster to use, but the clear majority of the respondents regarded the sunburst chart version as their favorite prototype. The two experts who participated in the expert review had different opinions on which one of the prototype versions they would select to be utilized in practice. Concerning the prototypes by the artificial hierarchy: These prototypes also received positive feedback, but the possibility to change the order of features in the hierarchy was considered as an extremely important development idea. ACM Computing Classification System (CCS): Human-Centered Computing → Visualization → Visualization Techniques Human-Centered Computing → Visualization → Empirical Studies in Visualization