基于离散Bass模型的双十一销售预测

The Double Eleven Sales Forecast Based on Exact Discrete Bass Model

阿里巴巴双十一购物节的销售额走势,不仅是相关企业规划决策的依据,也是我国经济活动的重要方向性指标,运用Bass模型预测双十一销售额具有十分重要的意义.对普通正向的离散Bass模型采用普通最小二乘法得到的参数估计值偏离于原连续模型参数的真实值,因为存在着多重共线性和时间区间偏差.为克服这两个缺点,根据一类黎卡提方程构建精确的离散Bass模型.研究显示该模型及其解均收敛于原连续模型.给出基于非线性最小二乘法则的搜索算法,估计模型参数,拟合效果较好.最后将模型结果运用到阿里巴巴双十一的销售额预测,结果表明2018销售额将迎来拐点,增速将放缓.

The future tendency of Alibaba Double Eleven shopping concerns not only the related enterprises to make planning and decision, but also involves important directional leading indicator for economic activity of our country. It is significant to apply Bass model to forecast the Double Eleven sales. However, for ordinary forward discrete Bass model, parameter estimation obtained by using the ordinary least squares procedure may deviate from the true value due to multicollinearity and the time-interval bias. To overcome the two shortcoming, a discrete exact Bass model followed discrete Riccati Equation is constructed. Research shows that the discrete exact Bass model and its' solution converge almost surely to the original continuous model. This paper builds a nonlinear least squares- based search algorithm to estimate the parameters, which produces a good fitting effect. Then it is applied to forecast the Alibaba Double Eleven shopping sales and the results indicate that sales will meet a turning point in 2018 and then will slow down gradually.