An Improved Demand Forecasting Model Using Deep Learning Approach and Proposed Decision Integration Strategy for Supply Chain

Table 1

Time series algorithms used in demand forecasting.

Models	Formulation	Definition of Variables
Regression Model 1		Variability in weekly customer demand; special days and holidays (X1), discount dates (X2), days when the product is closed for sale (X3), and sales (X4) that cannot explain these three activities but are exceptional.
Regression Model 2		R1 is modeled by adding weekly partial-regressive terms .
Regression Model 3		The R3 model adds the shadow regression terms to the R1 model for each month of the year , and for the months of the week end result.
SVR (Support Vector Regression)		The SVR prediction is used for forecasting by using inputs X1, X2, X3 mentioned regression model 1, each month of the year , and the months of the week .
Exponential Smoothing Model		Exponential smoothing is generally used for products where there is no trend or seasonality in the model.
Holt-Trend Methods		Additive Holt-Trend model has been evaluated for products with level (L) and trend (T).
Holt-Winters Seasonal Models		Additive Holt-Winters Seasonal models have been evaluated for products with level (L), seasonality (S) and trends (T)
Two-Level Model 1		In this method, R1 model was first applied to the time series. Later, the demand values estimated by this model are subtracted from the customer demand data and the residual values are fitted with exponential correction and Holt-Winters Exponential Smoothing Model. Estimates of these two metrics were collected and the product demand estimate was calculated and evaluated.
Two-Level Model 2		In this method, R1 model was first applied to the time series. Later, the demand values estimated by this model are subtracted from the customer demand data and the residual values are fitted with exponential correction and Holt-Winters Trend Model. Estimates of these two metrics were collected and the product demand estimate was calculated and evaluated.
Two-Level Model 3		In this method, R1 model was first applied to the time series. Later, the demand values estimated by this model are subtracted from the customer demand data and the residual values are fitted with exponential correction and Holt-Winters Seasonal Model. Estimates of these two metrics were collected and the product demand estimate was calculated and evaluated.