Multivariable Linear Regression Model for Promotional Forecasting:The Coca Cola - Morrisons Case
Zheng, Yiwei/Y (2009) Multivariable Linear Regression Model for Promotional Forecasting:The Coca Cola - Morrisons Case. [Dissertation (University of Nottingham only)] (Unpublished)
This paper describes a promotional forecasting model, built by linear regression module in Microsoft Excel. It intends to provide quick and reliable forecasts with a moderate credit and to assist the CPFR between the Coca Cola Enterprises (CCE) and the Morrisons. The model is derived from previous researches and literature review on CPFR, promotion, forecasting and modelling. It is designed as a multivariable linear regression model, which involves several promotional mix as variables including percentage discount, display, and holidays. Before modelling, all data and variables have been tested for their validity by two tests: the trend test and the up/downlift-average test. The model has also been conducted twice: the first time is to use a part of the data to define the structure of the model and the second time is to use all the data to finalize the model by deciding its coefficients. The model is capable to make forecast for 26 products and to forecast for several new promotions. The performance of this model is satisfactory in terms of the adjusted R2 (over 80%) and the MAPE (lower than 20%). A user-friendly interface is also provided to facilitate the use of the model in the actual forecasting. However, the model can be further improved both from the modelling method and the variable refining.
Actions (Archive Staff Only)