Visualizing Missing Values

Many real world data sets have data items with missing values. Values can be missing for many different reasons, such as sensor failure, respondents forgetting or refusing to answer a question in a survey, or a certain feature not being applicable to certain subsets of data. When visualizing data, some visualizations can easily handle missing values, while for others it is not obvious how to represent them without the resulting visualization being misleading. We give examples of different ways our system for interactive visual exploration of data handles missing data. These examples come from real world big data projects we took part in. Different ways to visualize missing values work well with different visualizations. Coordinated multiple views is a powerful way to visualize data with missing values, and having several views of the data helps explore the properties of the items with missing values.