Abstract
Linear regression is often used to analyse and summarize data, and to uncover, clarify, and simplify a data structure. The outcome of these activities depends on both the analyst's domain-specific knowledge and on the data. Analysing the data also affects the analyst's understanding about the data and, hence, the act of analysing data is inherently a recursive activity, with each new iteration potentially providing additional insights. This process calls for a strategy of exploratory data analysis that consists of techniques for flexibly analysing, summarizing, and re-expressing the data.