Practical Predictive Analytics

The Box-Cox Transform

We will now illustrate the Box-Cox Power Transformation (also known as Power Transformation). This is a generic transform that will search for optimal transformations to use on your data. This transformation optimizes an exponent called lambda, which is then applied to your data. It does this by iterating over all exponents between -5 and +5 until it has found the best one to transform your data to a normal distribution:

boxcox(x~1)

As you can see from the plot produced from the boxcox() function (the third plot), the optimal exponent to apply the data will be somewhere between 0 and 1.

Next, we will apply the power function. The powerTransform() function will apply the optimal lambda just calculated to the original ...