Comparative Evaluation of an Interactive Time‐Series Visualization that Combines Quantitative Data with Qualitative Abstractions

Abstract

In many application areas, analysts have to make sense of large volumes of multivariate time-series data. Explorative analysis of this kind of data is often difficult and overwhelming at the level of raw data. Temporal data abstraction reduces data complexity by deriving qualitative statements that reflect domain-specific key characteristics. Visual representations of abstractions and raw data together with appropriate interaction methods can support analysts in making their data easier to understand. Such a visualization technique that applies smooth semantic zooming has been developed in the context of patient data analysis. However, no empirical evidence on its effectiveness and efficiency is available. In this paper, we aim to fill this gap by reporting on a controlled experiment that compares this technique with another visualization method used in the well-known KNAVE-II framework. Both methods integrate quantitative data with qualitative abstractions whereas the first one uses a composite representation with color-coding to display the qualitative data and spatial position coding for the quantitative data. The second technique uses juxtaposed representations for quantitative and qualitative data with spatial position coding for both. Results show that the test persons using the composite representation were generally faster, particularly for more complex tasks that involve quantitative values as well as qualitative abstractions.

Qualizon graphs

“…Two suitable candidates have been already introduced in Section 2: KNAVE-II [27] and STZ [6]. A comparative evaluation [4] has already shown that STZ is as efficient as KNAVE-II in terms of completion time and error rate, and is even better for more complex tasks. For this reason, we consider STZ as a more appropriate candidate.…”

“…At the maximum height, the time series is visualized as a line plot with coloured annotations highlighting transitions between qualitative abstractions; at the minimum height, quantitative data disappear and qualitative abstractions are visualized as coloured rectangles. A comparative evaluation between this hybrid visualization, also known as SemanticTimeZoom (STZ), and KNAVE-II is presented in [4]. That user study, a task-based controlled experiment with 20 participants, revealed that STZ outperforms KNAVE-II in terms of task completion time, while there is no significant difference in terms of error rate.…”

Visual Analysis of Compliance with Clinical Guidelines

“…VIE-VENT uses temporal abstractions adapted to the context and derives high level trend patterns [19]. Among systems that apply abstractions especially for visualization, are Knave-II [27] and Midgaard [4,5]. Midgaard combines patient observation and treatment (plan) data, offers mechanism for semantic zooming, including data abstraction mapped to color in the lowest level of detail.…”

Interactive Visual Transformation for Symbolic Representation of Time-Oriented Data

“…STZ is a visualization method for qualitative abstractions and the associated quantitative time-oriented data. This interactive visualization technique, which is referred to as SemanticTimeZoom (STZ), is adopted from the Midgaard system [38,39]. While, in previous publications the qualitative abstractions were defined on basis of a single variable, typically based on severity ranges from domain knowledge, we can apply here the event data from our classification as common abstraction for all variables.…”

“…The description is rated as Figure 11. The SemanticTimeZoom (STZ) adopted from Midgaard [38,39] can interactively combine numerical raw data with qualitative events.…”