Authors:
Nicolas Fuchs
;
Pierre Vanhulst
;
Raphaël Tuor
and
Denis Lalanne
Affiliation:
Human-IST Institute, Université of Fribourg, Boulevard de Pérolles 90, Fribourg, Switzerland
Keyword(s):
Human-Centered Computing, Visualization Design and Techniques, Evaluation Methods.
Abstract:
Is it possible to visualize complex Origin-Destination (OD) data along with relevant spatio-temporal data?
In this paper, we tackle this issue by presenting Flowstrates++, an augmented version of Flowstrates which
aims to visualize additional time-series datasets linked with OD data. On top of Flowstrates’ heatmap, we
designed a second heatmap for spatio-temporal data, synchronized on the temporal axis, as well as other
dataset comparison features. Two versions of Flowstrates++ have been designed and implemented: Switch,
that displays one external dataset at a time, and Combi (for ”combined”), that displays two external datasets
at the same time. We aimed to assess to which extent both variants spur users into making multidimensional
findings. To achieve this goal, we evaluated both variants with ninety participants: ten were pilot users in
live remote sessions, and eighty were provided by Prolific.co, a crowd-sourcing platform. In a within-groups
study, these participants w
ere asked to take relevant annotations about the data on both variants, and to evaluate
them through a survey. We then classified the annotations using a framework whose validity was evaluated
with an Intercoder Agreement and Fleiss’ Kappa. We found that the Combi variant yielded consistently better
results, both in terms of number of produced multidimensional annotations, and in terms of appreciation of the
participants. Yet regardless of the variant, our solution allows users to highlight potential correlations between
time-series data and temporal OD data.
(More)