ABSTRACT

The data set consists of selected characteristics of 74 automobile models sold in the United States (US) in 1979. The variables can be roughly categorized into three groups: performance, size, and price. In working with the automobile data, the author encounters several of the complications that are common in practice. There are missing values, and there are different measurement scales. Interestingly, in terms of weight and price, three of the four most expensive US cars are Cadillac Seville, Cadillac Eldorado, and Lincoln Versailles fall in line with the foreign cars. The order of assigning variables to aspects of a symbol can have a major effect on the appearance of the collection of symbols and on the relationships in the data that the plot brings out. The draftsman's display for three dimensions generalizes easily to applications involving multivariate data. The division of points into two branches, or arms, in the plot is apparent, even at this scale.