Forecasting Campaign Sales Using Artificial Neural Networks

Abstract: This thesis will investigate the characteristics and possibilities of implementing a forecasting model for campaign sales in Coop stores. The implementation will be done by the use of multiple linear regression and artificial neural networks. Advantages and drawbacks of these models will be investigated and discussed. The campaign characteristics will be evaluated using linear methods, and several prediction models, of varying complexity, will be proposed and tested on real sales data. It is concluded that while the complex models perform better, the tradeoff between loss in accuracy and increased simplicity, also making the models more generic, may justify the use of the simpler models.