Time series data may be stationary or non-stationary in nature. Stationary implies flat without periodic fluctuations, while non-stationary data typically has frequent shifts in value. You see time series analysis generally used for non-stationary data, such as evaluating and predicting retail sales. In this example, we will again utilize a study exercise (available on GitHub) to demonstrate the fundamental steps involved in the analysis and forecasting of retail sales data, as implemented with IBM Watson Studio.