Understanding and preparing chart data

Chart design depends on more than the values stored in a dataset. We need to understand what each observation represents, how the observations relate to one another, and which comparison or pattern the chart should reveal. These properties determine both a suitable visual representation and the form the data needs to take.

Relationships among observations give datasets recognizable structures. This chapter focuses on four that recur in chart design: time-series data, categorical and proportional data, distributions, and spatial observations. This classification is neither exhaustive nor mutually exclusive, and a single dataset may contain several structures at once.

Each section begins by examining how one of these structures affects the interpretation of the data. It then considers which visual representations preserve that structure and how to prepare the values they require. Following this progression helps us recognize which properties matter to the intended comparison, choose an appropriate chart, and prepare the required values without losing their meaning.