Patterns and Sequences: Interactive Exploration of Clickstreams to Understand Common Visitor Paths.

Area of Science:

• Data Science
• Human-Computer Interaction
• Web Analytics

Background:

• Web clickstream data is characterized by long, high-dimensional sequences of multivariate events, posing significant analytical challenges.
• Effective visualization requires presenting information at multiple levels of granularity, enabling seamless user navigation.

Purpose of the Study:

• To develop an analytical pipeline for understanding complex web clickstream data.
• To propose methods for reducing the complexity of extracted patterns.
• To inform the design of visual interfaces for exploring sequential data.

Main Methods:

• Identification of four key levels of granularity: patterns, segments, sequences, and events.
• An analytic pipeline involving pattern mining, pattern pruning, and coordinated exploration.
• Discussion of maximal sequential patterns and visualization design considerations.

Main Results:

• Demonstrated the viability of the proposed analytical pipeline through a case study.
• Proposed methods for pattern reduction and efficient visualization of sequential data.
• Gathered user feedback to assess the strengths and limitations of the approach.

Conclusions:

• The multi-level granularity approach and visualization pipeline effectively address the complexity of web clickstream data analysis.
• The proposed methods facilitate easier navigation and understanding of sequential patterns within user navigation data.
• Further refinement based on user feedback can enhance the practical application of these visualization techniques.