Abstract
When exploring time series datasets, analysts often pose "which and when" questions. For example, with world life expectancy data over one hundred years, they may inquire about the top 10 countries in life expectancy and the time period when they achieved this status, or which countries have had longer life expectancy than Ireland and when. This paper proposes TimePool, a new visualization prototype, to address this need for univariate time series analysis. It allows users to construct interactive "which and when" queries and visually explore the results for insights.
Introduction
Analyzing univariate time series data—where a single variable is tracked over time for many different entities—is a fundamental task in fields ranging from public health to finance. However, as the number of entities grows (e.g., tracking life expectancy for 200 countries over a century), standard line charts suffer from the "spaghetti chart" problem. The resulting visual clutter makes it nearly impossible to track individual trajectories or identify transient patterns.
TimePool addresses a specific gap in current visualization tools: the difficulty of answering "Which and When" questions. These questions require identifying which entities meet specific criteria and exactly when those conditions occurred. In a standard line chart, overlapping and crossing lines obscure these answers. TimePool solves this by combining traditional line charts with interactive, juxtaposed timelines.
The TimePool Interface
TimePool introduces a multi-view interface designed to facilitate an exploratory loop. The system consists of four primary components:
- Line Chart (Top Left): This view displays the raw data. Its primary innovation is direct manipulation; users can drag threshold lines or curves across the chart to define queries visually.
- Detail View (Bottom Left): To eliminate clutter, TimePool moves from an "overlay" to a "juxtaposed" view. This view provides a set of timelines (one row per entity). When an entity meets the query criteria, its segment in the timeline is colored (e.g., green for "high," red for "low"); otherwise, it remains gray.
- Overview (Bottom Right): A zoomed-out version of the detail view that allows analysts to identify global patterns and temporal clusters across the entire dataset at a glance.
- Control Panel (Top Right): Provides fine-grained control over query parameters and data transformations.
Key Innovations in Querying
Beyond simple value thresholds, TimePool supports dynamic queries based on derived values, including ranking, percentage change, net change, and variance. This allows for sophisticated analysis, such as identifying which countries experienced the highest variance in GDP during a specific decade.
To further refine the analysis, TimePool includes:
- Temporal Sorting: Users can sort timelines by the total duration a condition was met. This distinguishes between entities that are "consistently" in a state versus those that are only "transiently" in that state.
- Grouping: Entities can be grouped by categories (e.g., continents) to reveal regional trends, such as the unique challenges faced by Sub-Saharan Africa in the 1990s.
- Ego Mode: A comparison tool where a single entity (the "ego") serves as a dynamic baseline. All other entities are then visualized relative to the ego's performance.
Evaluation and Results
The researchers evaluated TimePool using datasets including World Life Expectancy (200 countries, 1900–2012), World GDP, and Beijing Air Quality. A user study compared TimePool to a baseline system consisting of a standard line chart with query tools.
- Efficiency: For "Extreme" tasks (e.g., finding top members), TimePool users were nearly 3x faster, completing tasks in 43 seconds compared to 124 seconds for the baseline.
- Accuracy: TimePool significantly improved accuracy. For comparison tasks, TimePool users achieved 96% accuracy, while baseline users achieved only 63%.
- Expert Feedback: 18 domain experts rated the system highly (averaging ~8.9/10 for usefulness and intuitiveness). Experts noted that the system provides an intuitive way to uncover patterns hidden in complex time series datasets.
Discussion and Future Work
While TimePool significantly improves the analysis of univariate time series, some limitations remain. Experts noted a slight initial learning curve to master the coordination between views. Additionally, while the system scales better than line charts, vertical scrolling may become a bottleneck for datasets containing thousands of entities.
Future work will focus on extending the "Which and When" logic to multivariate data—allowing users to query across multiple variables simultaneously (e.g., "Which countries had high GDP and low infant mortality?"). The researchers also aim to integrate more diverse data sources and automated annotation features to further assist the exploratory process.
Summary updated