In Example 5.9, the design matrix X has an intercept column, and covariate columns xi, and x2 i . Rescaling the covariate columns to xi/x and (xi/x)2 drops the condition substantially, suggesting the...

In Example 5.9, the design matrix X has an intercept column, and covariate columns xi, and x2 i . Rescaling the covariate columns to xi/x and (xi/x)2 drops the condition substantially, suggesting the condition number for X may be artificially large. Consider the situation where the change in the design matrix arose from changes in the recording date, say, moved one day because of a holiday (1/365.25), or due to a leap year effect (.25/365.25). What would be the effect on the solution to this least squares problem for changes such as these on the four design matrices given in Example 5.9?