Computing regression parameters
Once we've determined that two variables have some kind of relationship, the next step is to determine a way to estimate the dependent variable from the value of the independent variable. With most real-world data, there are a number of small factors that will lead to random variation around a central trend. We'll be estimating a relationship that minimizes these errors, striving for a close fit.
In the simplest cases, the relationship between variables is linear. When we plot the data points, they will tend to cluster around a line. In other cases, we can adjust one of the variables by computing a logarithm or raising it to a power to create a linear model. In more extreme cases, a polynomial is required. The process of linear regression estimates a line that will fit the data with the fewest errors.
In this recipe, we'll show how to compute the linear regression parameters between two variables. This will be based on the...
