INTRODUCTION TO REGRESSION

[Link Map]
begin comments text sound
line

Regression is a technique for describing, testing and estimating the relationship between a (random) variable and another variable (or variables) which may be random or controlled. The relationship is described by specifying the expected value of the dependent (or response) variable in terms of the other (explanatory) variable(s). The simplest type of relationship is a straight line which can be estimated by simple linear regression.

Read in the carcass data. This is data on 37 sheep carcasses which were visually graded for fatness, then had backfat depth measured by ultrasound and finally the actual fat content was determined by disection.

A scatter plot should always be examined before performing regression. Plot percent against depth using the xy scatter option of the chart wizard. Remember to skip the row with the missing depth value when selecting the data range. You should obtain the plot on the right.

This plot does show a reasonably linear relationship so it is justified to try linear regression.

line
Basic statistics in Excel   23.2.99   Page: 23 of 25