Dynamic low-dimensional linear projections of multivariate data collectively known as tours provide an important tool for exploring multivariate data and models. The R package tourr provides functions for several types of tours: grand, guided, little, local and frozen. Each of these can be viewed dynamically, or saved into a data object for animation. This paper describes a new package, spinifex, which provides a manual tour of multivariate data where the projection coefficient of a single variable is controlled. The variable is rotated fully into the projection, or completely out of the projection. The resulting sequence of projections can be displayed as an animation, with functions from either the plotly or gganimate packages. By varying the coefficient of a single variable, it is possible to explore the sensitivity of structure in the projection to that variable. This is particularly useful when used with a projection pursuit guided tour to simplify and understand the solution. The use of the manual tour is applied particle physics data to illustrate the sensitivity of structure in a projection to specific variable contributions.

This article discusses a high-dimensional visualization technique called the tour, which can be used to view data in more than three dimensions. We review the theory and history behind the technique, as well as modern software developments and applications of the tour that are being found across the sciences and machine learning. This article is categorized under: Statistical and Graphical Methods of Data Analysis > Analysis of High Dimensional Data Statistical and Graphical Methods of Data Analysis > Statistical Graphics and Visualization Statistical Learning and Exploratory Methods of the Data Sciences > Exploratory Data Analysis K E Y W O R D S data science, data visualization, exploratory data analysis, high-dimensional data, tours 1 | INTRODUCTION Data commonly arrive with more than two measured variables, which makes it more complicated to plot on a page. With multiple variables, especially if there is some association between variables, this would be called multivariate or high-dimensional data. When the variables are all numeric, or quantitative, visualization often relies on some form of dimension reduction. This can be done by taking linear projections, for example, principal component analysis (PCA; Hotelling, 1933), linear discriminant analysis (LDA; Fisher, 1936), or projection pursuit (PP; Friedman & Tukey, 1974). It is also common to reduce dimension with nonlinear techniques like multidimensional scaling (MDS; Kruskal, 1964), t-Distributed Stochastic Neighbor Embedding (t-SNE; van der Maaten & Hinton, 2008), or Uniform Manifold Alignment and Projection (UMAP; McInnes et al., 2020).The term "high-dimensional" here refers to the dimensionality of the Euclidean space. Figure 1 shows a way to imagine this. It shows a sequence of cube wireframes, ranging from one-dimensional (1D) through to five-dimensional (5D), where beyond 2D is a linear projection of the cube. As the dimension increases, a new orthogonal axis is added.

No abstract

Background: Academic performance is at the heart of hiring decisions and funding applications. It is based on a combination of qualitative and quantitative metrics. One of those is the venue in which scholarly publications are published. Depending on the perceived (qualitative) or measured (quantitative) prestige associated with a venue, a specific publication will have more or less weight. Objectives: We want to understand how visualization researchers consider the prestige of a venue when looking for papers that they could use in their own manuscripts, and how they determine the prestige of any given venue.Method: We ran an online survey open for 10 days that we sent out to visualization researchers.Results: We gathered 46 responses from a sample of convenience. We found that publication venue plays the biggest part in how visualization researchers assess research articles. Interestingly, rating systems and metrics are least important criteria for researchers when assessing the quality of a venue. Conclusion: We highlight the potential risks around focusing on venue when assessing research articles. We further underline the necessity to discuss with the community on strategies to switch the focus to robustness and reliability to foster better practices and less stressful publishing expectations.Reproducibility: Data, materials and preregistration available on osf.io/ch6p4/