To Estimate Parameters of Diffusion Theory Model with Genetic Algorithms--Taking the Diffusion of Mobile Subscribes in China for Example

Bass model have been used widely for diffusion curve of a new product in some countries of the world, but only a few studies have explored in the parameters´ estimation of bass model. In this paper, Bass model is adopted to fit the diffusion curve of mobile subscribers of China with different parameter-estimation techniques respectively, such as NLS(Non-Linear Least Squares ) ,OLS(Ordinary Least Squares), MLE(Maximum Likelihood Estimation) and GAs (Genetic Algorithms). The parameter estimates obtained from GAs exhibit good consistency comparable to NLS, OLS, MLE When enough time series data points are available. But when the datasets just own a few data points especially not including the peak sales time period , GA can provides better predictions of future sales than the other parameter-estimation techniques. In order to more effectively make forecasts in the mobile subscribes in China during the growth phase of a new product life cycle, we can use GA to estimate the parameters of bass model as compared to currently available estimation techniques, and then give the predictions of future mobile subscribers in China.