I. Introduction

As a result of social media, the widespread application of sensors, simulations, and the Internet, the amount of data produced is growing exponentially. This growth affects all domains, such as medicine, environmental science, physics, the social sciences, and economics, and has ushered in the era of big data. However, for data to be useful, it must be transformed into information that will facilitate the development of new insights. The ability to analyze, interpret, and gain insights from large amounts of data has not kept pace with the rapid technological advances in data acquisition, storage, management, and database techniques [5]. Consequently, new data analytics, computational, and statistical methods are being actively investigated to narrow this gap [10], [13]. As a crucial adjunct to these computational advances, visual analytics (VA) aims to facilitate the interpretation of massive or complex data by integrating interactive visualizations with human judgment and qualitative and quantitative analysis capabilities [5]. VA also uses the human ability to visually detect trends and patterns, assess anomalous conditions, and uncover complexities in the data that require further investigation with computational data analysis approaches [28]. Therefore, the human-in-the-loop paradigm is vital to the success of any analytics approach [14]. VA is facilitating the interpretation of big data but is also relevant for multidimensional data and data resulting from systems with many parameters. A particularly challenging problem is VA of time series data for determining meaningful temporal and statistical patterns [9], [24]–[26], [28].