Observation-Level and Parametric Interaction for High-Dimensional Data Analysis

Abstract

Exploring high-dimensional data is challenging. Dimension reduction algorithms, such as weighted multidimensional scaling, support data exploration by projecting datasets to two dimensions for visualization. These projections can be explored through parametric interaction, tweaking underlying parameterizations, and observation-level interaction, directly interacting with the points within the projection. In this article, we present the results of a controlled usability study determining the differences, advantages, and drawbacks among parametric interaction, observation-level interaction, and their combination. The study assesses both interaction technique effects on domain-specific high-dimensional data analyses performed by non-experts of statistical algorithms. This study is performed using Andromeda, a tool that enables both parametric and observation-level interaction to provide in-depth data exploration. The results indicate that the two forms of interaction serve different, but complementary, purposes in gaining insight through steerable dimension reduction algorithms. CCS Concepts: • Human-centered computing → Empirical studies in visualization;

Interactive Clustering

“…The concept of different interaction stages is not limited to clustering, of course. Our structure of this dimension matches quite well with the types of interaction identified by Self et al [87]. They distinguish "Parametric interaction" and "Observation-level interaction," with the former referring to users directly specifying or modifying design parameters of an algorithm, and the latter allowing users to interact with individual data items, usually in an interactive graphical tool.…”

“…They distinguish "Parametric interaction" and "Observation-level interaction," with the former referring to users directly specifying or modifying design parameters of an algorithm, and the latter allowing users to interact with individual data items, usually in an interactive graphical tool. Self et al [87] argue that these two forms of interaction offer distinct and complementary capabilities, and are likely to lead to different types of insights. Parametric interaction offers high degree of control but requires deep understanding of the analytical model; while observation-level interaction offers familiar interface embedded in the domain semantics, but changes made by the user may be incorrectly translated into model updates.…”

Making Sense of Scientific Simulation Ensembles With Semantic Interaction

“…‘magnets’ representing keywords), not data itself. On the other hand, tools such as ForceSPIRE [EFN12], Dis‐function [BLBC12], Andromeda [SH] and StarSPIRE [BNH14] focus on using human cognition to steer the underlying computations by directly manipulating the spatializations, giving users the chance to interact with data points and translate this feedback through a dimension‐reduction algorithm to a new view reflecting the user's interaction. This helps provide an intuitive space for strengthening insight creation and data understanding.…”

Bridging the gap between user intention and model parameters for human-in-the-loop data analytics

“…To study OLI, we designed and developed an interactive interface, Andromeda (Figure 1), that visualizes high-dimensional data using WMDS [8]. Andromeda's object view (Figure 1a) visualizes the WMDS projection and the parameter view ( Figure 1b) displays the weights as horizontal lines.…”

“…Previous research involving a controlled user study found the interactions within Andromeda allowed users to perform successful data analyses [8]. OLI and parametric interaction provide the user with two analysis angles: object-centric and dimension-centric.…”