Evolutionary Visual Exploration: Evaluation With Expert Users

Abstract

We present an Evolutionary Visual Exploration (EVE) system that combines visual analytics with stochastic optimisation to aid the exploration of multidimensional datasets characterised by a large number of possible views or projections. Starting from dimensions whose values are automatically calculated by a PCA, an interactive evolutionary algorithm progressively builds (or evolves) non-trivial viewpoints in the form of linear and non-linear dimension combinations, to help users discover new interesting views and relationships in their data. The criteria for evolving new dimensions is not known a priori and are partially specified by the user via an interactive interface: (i) The user selects views with meaningful or interesting visual patterns and provides a satisfaction score. (ii) The system calibrates a fitness function (optimised by the evolutionary algorithm) to take into account the user input, and then calculates new views. Our method leverages automatic tools to detect interesting visual features and human interpretation to derive meaning, validate the findings and guide the exploration without having to grasp advanced statistical concepts. To validate our method, we built a prototype tool (EvoGraphDice) as an extension of an existing scatterplot matrix inspection tool, and conducted an observational study with five domain experts. Our results show that EvoGraphDice can help users quantify qualitative hypotheses and try out different scenarios to dynamically transform their data. Importantly, it allowed our experts to think laterally, better formulate their research questions and build new hypotheses for further investigation.

Statsplorer

“…For correlation questions, EvoGraphDice uses an interactive scatterplot matrix generated by evolutionary algorithm to help users discover complex relationships in multi-dimensional data [4]. For research questions involving differences, we discuss the existing tools below.…”

A Five-Level Design Framework for Bicluster Visualizations

“…ForceSPIRE [24] is an example of such interactions (e.g., moving, annotating and highlighting) that re-weight the term dimensions in the distance metric and recalculate similarities among documents. EvoGraphDice [11,14] employed a similar idea to dynamically change a scatterplot matrix. Object-LI enable users to focus on the data and perform exploratory analysis (for Bicluster-LR and Chain-LR ) by implicitly tuning the algorithm parameters.…”

Evolutionary visual exploration

“…Recent extensions, described in [3], are (i) a Genetic Programming (GP) algorithm allowing the manipulation of nonlinear combinations of dimensions as variable size mathematical formulae, (ii) user assessment of proposed views is explicitly captured via a slider, (iii) a surrogate function based on some specific geometric measurements (scagnos- Figure 2: Nine scagnostics measures from [16].…”

“…In [3], we validated our general approach of combining visual analytics with stochastic optimisation to aid data exploration, by conducting an observational study with expert users from various domains. Our results showed that EvoGraphDice can help users quantify qualitative hypotheses and try out different scenarios to dynamically transform their data.…”

“…EvoGraphDice [3,4] is an EVE prototype for the exploration of multidimensional datasets: the system proposes interesting views based on both objective measures, such as visual patterns in the two-dimensional projections of the data, and subjective measures corresponding to user satisfaction with the presented view. EvoGraphDice was built as an extension of an existing scatterplot matrix inspection tool.…”