İşlemsel zeka ve ATA metot kullanılarak kesikli taleplerin zaman serisi tahminlemesi

Intermittent demand forecasting is substantially important for firms and commercial activities. Recently, many researchers have focused on the methods which are used to forecast for intermittent demand and developed varied forecasting techniques. The outstanding method among these developed techniques is Croston method which is developed by inspired from exponential smoothing methods. Primarily, exponential smoothing (ES) method which is universally used for more than 50 years is still one of the most prevalent forecasting techniques. Moreover, this model's superiority and success is proved by the famous forecasting competition (M-competitions). Despite its supremacy and widespread use in many areas, ES models have some drawbacks such as the initial value problem that negatively affect the accuracy of forecasts. On the other hand, Croston method is widely used in forecasting of intermittent demand and inventory (stock) control. Since these demands usually include zero values, using the groundbreaking method developed by Croston in this data becomes inevitable. Nevertheless, there are some shortcomings proved such as biasedness in this method. In this study, our aim is to compare the consequences of combining Croston and ATA method as new method for an alternative of exponential smoothing and the other goal is to adjust ATA method's properties to Croston method. Hence, ATA method will be used to intermittent demand forecasting and also proposed this method will be compared statistically to all mentioned methods.