This document describes methods for determining whether a time series reflects linear or nonlinear processes. It compares the forecasting performance of linear versus nonlinear models. The two-step procedure involves using simplex projection to identify the best embedding dimension, and then using that dimension in the S-map procedure to assess nonlinearity. Both methods evaluate out-of-sample forecast skill by dividing the time series in half, using one half to build the model and the other to evaluate forecasts.