About the Data Smoothing Method

Data smoothing techniques eliminate noise and extract real trends and patterns. The equations below are used to calculate the forecasts during the historical periods.

Data Smoothing Methods

Method	Description
Random	This method is best used when each period's data has no relationship to the pattern in the previous data. Under this condition, the best prediction for the next value in a series is the average of all previous data points.
Random Walk (Naïve)	A random walk exists if the next data point is equal to the last data point plus some random deviation. Many financial securities move in this manner. Under this condition, the best prediction for the next value in a series is the last value.
Moving Average	This method works well if the data contains no trend. User-supplied parameters: - n: Number of consecutive points to average The number of consecutive points to average must be greater than zero. Higher values cause greater smoothing.
Simple Exponential Smoothing	This method works well if the data contains no trend or cyclic pattern, and the most recent data points are more significant than earlier points. User-supplied parameters: - a: Smoothing factor The smoothing factor must be between zero and one inclusive.
Holt's Linear Method	This method works well if the data contains a trend but no cyclic pattern. It is also known as linear exponential smoothing or double exponential smoothing. User-supplied parameters: - a: Level smoothing factor - b: Trend smoothing factor Both the level and trend smoothing factors must be between zero and one inclusive.
Winter's Multiplicative Season	This method works well if the data contains a trend and a cyclic pattern. It is also known as Holt-Winter’s method with multiplicative seasonality, seasonal exponential smoothing, or triple exponential smoothing. User-supplied parameters: - a: Level smoothing factor - b: Trend smoothing factor - c: Seasonal smoothing factor - p: Season period All smoothing factors must be between zero and one inclusive. The season period defines the number of periods occurring between two successive starts of a season. The season period must be greater than zero.
Winter's Additive Season	This method works well if the data contains a trend and a cyclic pattern. It is also known as Holt-Winter’s method with additive seasonality, seasonal exponential smoothing, or triple exponential smoothing. User-supplied parameters: - a: Level smoothing factor - b: Trend smoothing factor - c: Seasonal smoothing factor - p: Season period All smoothing factors must be between zero and one inclusive. The season period defines the number of periods occurring between two successive starts of a season. The season period must be greater than zero.
Croston's Intermittent Demand	Some products are sold intermittently, making it difficult to forecast their demand. You may not have these products in stock when a customer requests them, or you may stock these products continuously and suffer from excess inventory. Croston's intermittent demand is useful when historical data is distributed randomly and sporadically, or is independent of a demand interval. The method creates exponential smoothing estimates of the average interval between demands and the average size of each demand. Essentially, the method calculates the probability of positive demand in a period and the likely value of the demand.
ARIMA (Box-Jenkins)	This method has both manual and automatic modes. In manual mode, you can configure several settings: - Autoregressive terms (0+) - Differencing order (0,1,2) - Moving average terms (0,1,2) - Period (season.cycle)