The Dataset


The dataset we will use here is Penrose et al. (1985).  This dataset contains 252 observations and 19 variables, and is described below. First we need to load the data.

 > fat <- read.table("fatdata.txt",header=TRUE, sep="\t")

> str(fat)

 Table 1. Description of Percent Body Fat data

Variable

Description

id           

Case Number

pctfat.brozek

Percent body fat using Brozek's equation, 457/Density - 414.2

pctfat.siri  

Percent body fat using Siri's equation, 495/Density - 450

density      

Density (gm/cm^3)

age         

Age (yrs)

weight       

Weight (lbs)

height       

Height (inches)

adiposity    

Adiposity index = Weight/Height^2 (kg/m^2)

fatfreeweight

Fat Free Weight = (1 - fraction of body fat) * Weight, using Brozek's formula (lbs)

neck        

Neck circumference (cm)

chest        

Chest circumference (cm)

abdomen      

Abdomen circumference (cm) "at the umbilicus and level with the iliac crest"

hip          

Hip circumference (cm)

thigh        

Thigh circumference (cm)

knee        

Knee circumference (cm)

ankle        

Ankle circumference (cm)

biceps       

Extended biceps circumference (cm)

forearm      

Forearm circumference (cm)

wrist

Wrist circumference (cm) "distal to the styloid processes"

The percentage of body fat is a measure to assess a person's health and is measured through an underwater weighing technique. In this lecture we will try to build a formula to predict an individual's body fat, based on variables in the dataset.