以中国移动用户的增长扩散情况为研究背景,通过选择不同的起始点、不同的时间间隔以及不同长度的数据序列来拟合BASS模型,然后比较不同条件下参数估算的结果.实验结果表明BASS模型对这些初始条件的改变很敏感,不同的时间序列对模型的参数估算影响很大;另外,BASS模型的两种不同曲线形式,S型累积曲线及钟形增长曲线对同一时间序列也有不同的拟合效果.S型累积曲线的拟合效果要好于钟形增长曲线.最后给出选取BASS模型的数据序列时应该注意的问题:数据序列初始点的数值应大于最大市场潜力与创新系数的乘积,时间间隔应为等距,数据序列的长度至少要7个点以上.%The paper uses the time serial datasets of the adopting mobile subscribers in China to fit the bass model with different starting points, different time intervals and different length of datasets. The results show that it is very important for the estimation of parameters of bass model in different starting condition of datasets, and the fitness of the cumulative curve of bass model is better than the growth curve of bass model, even though the two curves of bass model adopt the same starting conditions of datasets. The paper gives some advices on how to choose the datasets for the fitness of bass model: the value of starting point is bigger than the product of number of eventual adopters and coefficient of innovation, the time interval is same, and the length of dataset is over 7 points.
展开▼
