In Chapter 3 we learned how to find the LSRL for a set of data.  In Chapter four we will learn how to transform data in order to calculate a good prediction equation for a curved data set.

Here is an example of a scatterplot with a curved pattern.

This is the original data of a relationship.

Unlike in chapter 3, this data is curved.  Let's find the LSRL and see what happens.

We are going to have a strong correlation (r could be .91)

Can you tell from the picture above that the residual plot will show a pattern?  Remember that residuals = observed-predicted.

This is the residual plot for the LSRL in the second picture.

We can clearly see a pattern in our residual plot, meaning our LSRL is NOT a good fit for the data.  Simpily there is something else we can do to find a better equation to make predictions.

Correlation
Transforming
Transforming Exponential Relationships
Transforming Power Relationships
Transforming Logarithmic Relationships
Causation
Practice
Relations and Categorical Data