Getting Started: Exploratory Data Analysis of Tropical Cyclones

Creating a Box Plot

The data set contains several variables that measure the size of a tropical cyclone. One of these is the radius_eye variable, which contains the radius of a cyclone's eye in nautical miles. (The eye of a cyclone is a calm, relatively cloudless central region.) The radius_eye variable has many missing values, because not all storms have well-defined eyes.

In this section you create a box plot that shows how the radius of a cyclone's eye varies with the Saffir-Simpson category. The figures in this section assume that you have excluded observations with low wind speeds as described in the "Creating a Bar Chart" section.


Select Graph \blacktriangleright\,Box Plot from the main menu.

The box plot dialog box appears as in Figure 2.10.

ugedaboxdlg.png (8799 bytes)

Figure 2.10: Box Plot Dialog Box



Select the variable radius_eye, and click Set Y.

Select the variable category, and click Add X.

Click OK.

A box plot appears. Move the box plot so that it does not cover the data table or other plots.

The box plot summarizes the distribution of eye radii for each Saffir-Simpson category. The plot indicates that the median eye radius tends to increase with storm intensity for tropical storms, category 1, and category 2 hurricanes. Category 2 - 4 storms have similar distributions, while the most intense hurricanes (Cat5) in this data set tend to have eyes that are small and compact. The box plot also indicates considerable spread in the radii of eyes.

Recall that the radius_eye variable contains many missing values. The box plot displays only observations with nonmissing values, corresponding to storms with well-defined eyes. You might wonder what percentage of all storms of a given Saffir-Simpson intensity have well-defined eyes. You can determine this percentage by selecting all observations in the box plot and noting the proportion of observations that are selected in the bar chart.


Drag out a selection rectangle in the box plot around the category 1 storms.

In the bar chart in Figure 2.11, note that approximately 25% of the bar for category 1 storms is displayed as selected, meaning that approximately one quarter of the category 1 storms in this data set have nonmissing measurements for radius_eye.

ugedabarbox.png (64370 bytes)

Figure 2.11: Proportion of Category 1 Storms with Well-Defined Eyes



Drag the selection rectangle to select eye radii in other categories.

The selected observations displayed in the bar chart reveal the proportion of storms in each Saffir-Simpson category that have nonmissing values for radius_eye. Note in particular that very few tropical storms have eyes, whereas almost all category 4 and 5 storms have well-defined eyes.


Click outside the plot area in any plot to deselect all observations.

Previous Page | Next Page | Top of Page