2. Related Surveys

This section outlines closely related surveys in this field but have a different scope. See the work of McNabb and Laramee for a comprehensive overview of survey papers in information visualisation. Firstly, we provide an overview of the survey on visual analytics of financial data. Secondly, we describe Zhang et al.'s survey outlining state-of-the-art commercial visualisation systems readily available on the market.

Ko et al. presented a comprehensive survey on visual analytical approaches to financial data. This paper aims to classify financial visualisation papers as well as make contributions as to the requirements for a financial visualisation paper.

The survey first identifies financial task requirements by interviewing analysts with a background in financial analysis. They introduce metrics by which financial visualisations can be quantified. Secondly, the survey collects and classifies previously published research in the field of financial visualisation. The requirements for VA systems are a product of the interview process with industry experts. Ranging from R1 (basic requirements) to R7 (Maximised utility).

  1. R1: Provide sufficient information to deduce basic patterns including historical and context data.
  2. R2: Automated techniques for pattern detection, trends and anomalies.
  3. R3: User interaction with the system. Enabling data resolution selection (drill-down), and data comparison.
  4. R4: Statical analysis of trends and anomalies identifying "statistically significant" trends.
  5. R5: Forecasting for future trends based on currently available data.
  6. R6: Additional functions for data cleansing, customisation and presentation.
  7. R7: Clear visualisations that avoid occlusion as well as supporting R6 and R3 functionality.

Ko et al. highlighted the discrepancy in the volume of research between the classes of financial visualisation. The discrepancy is attributed to data privacy issues within the financial community.

Zhang et al. presented a comparison of the industry-leading visual analytics software used in Big Data Analysis. The Top 10 most prolific pieces of visual analytics software are compared. The four most popular of these are: Tableau, Spotfire, QlikView, and JMP.

Contributions include a comparison of how the software handles data, what frameworks are used, and the efficiency of the data management. Aspects such as data import ease, data compatibility, etc. are taken into account when assessing the software. Four subsections of criteria are used to assess the automatic analysis of the software: statistics, data modelling, dimensionality reduction, and visual query analysis. This is considered a way of learning about the data without greatly customised user input. Features such as pattern recognition are mentioned as useful in some of the software but not implemented across all of them. Visualisation techniques are divided into two subsections: graphical representations and interaction techniques. System and architecture is divided into stand-alone desktop applications and server-sided dashboard tools.

The visual analytics field was initially defined by Thomas and Cook, and then further refined by Keim et al. Using these as a basis for analysis, Zhang carried out the software comparison research. Our survey is neither financial visualisation nor a survey of software tools.