Study On Generalized Bass Model Based On Ant Colony Optimization

In order to seize market share, predatory excess profits and maintain competitive advantage, companies often use a variety of marketing methods to promote products rapid products diffusion. From a marketing perspective, advertising as an effective marketing tool plays an important role in enhancing performance, improving corporate image and culture, and flourishing market economy. Existing marketing survey shows that advertising as one of marketing means has an increasingly important influence on new product diffusion. Hence, the innovation diffusion model including advertising will be able to make the theoretical model more appropriate to actual situations, and then it helps managers to increase prediction accuracy of new product diffusion, grasp the latest market trends, control market structure and make decision consistent with the facts of the market.Innovation diffusion model is the important model that composing the study of simulating or forecasting new products marketing diffusion, in which Bass model is the most applying one, such as Eastman,Kodak,IBM,Sear and Hewlett-Packard use Bass Model to forecast new products marketing diffusion and sales, to provide technical support for developing their marketing strategies. Although the model has been successfully applied in various fields, but it also has some limitations, for example, it Bass model didn't include advertising. In fact, marketing has an increasingly important influence on new product diffusion and promoting proliferation of new products. Therefore, new product diffusion model including advertising more scientifically and reasonably reflects the true proliferation of new market rules.Therefore, based on Bass Model, the paper gives Generalized Bass Model that includes advertising. Parameter estimation methods of existing prediction model are largely confined to requiring enough data, and diffusion models which have analytical solutions and discrete form. Taking into consideration the complexity of the model, the paper presents a new method that estimates the parameters of new products diffusion model, which is Ant Colony Optimization. The algorithm is able to not only overcome the limitation of the existing algorithms, but also has the following advantages:1) positive feedback; 2) parallel distributed computing; 3) strong robustness, and easy to combine with other algorithms. Finally, as an example, the proliferation of mobile communication technology in China was used to demonstrate this algorithm. And Ant Colony Optimization,Ordinary Least Square,Maximum Likelihood Estimate and Nonlinear least squares were applied to the mobile communication technology respectively as the parameter estimation approaches and the results were then analyzed and compared. The results show that the estimated effect of ACO is better than others in the case of less historical data. At the same time, to compare empirical research results of Generalized Bass Model including advertising and Bass model, the results show that: Generalized Bass Model including advertising has better goodness of fit and forecast accuracy than other diffusion models.