Modelling and selecting among multiple time series methods

Businesses need a plan to efficiently use their resources, but an effective plan requires an accurate sales forecast. There are many time series forecasting methods, each appropriate for a certain type of data. Therefore, the selection of an appropriate method is critical for an accurate forecast. However, these methods have not been evaluated to determine which is most appropriate for the type of data used. Thus, a method is often selected without the user knowing whether it is appropriate for the situation. The current study evaluates several methods using simulated time series data, for example trend, seasonal, cyclical and irregular. The statistical error is calculated for each method using mean squared error, mean absolute percent error, mean absolute deviation and tracking. Based on the results, the Holt Winter method has the lowest error.