摘要
根据马尔可夫过程对汽车销售情况进行预测,提出采用颜色、排量及版本类型建立销售预测模型状态空间.进而根据某汽车销售门店的销售数据来分析状态空间的一步转移概率P(1),并结合初始概率S(0),对某时刻汽车各个状态的销售量所占比率进行预测;预测结果与实际销售数据进行比较,其绝对误差均小于5%.
The car sales forecast model with color, displacement and version was built according to the Markov process. The proportion of every state of car sales was forecasted according to the first-order transition probability P（1） combined with the initial probability S（0）. The absolute error between the actual sales and the forcasted sales was less than 5 %.
出处
《中国计量学院学报》
2015年第4期501-508,共8页
Journal of China Jiliang University
基金
浙江省自然科学基金资助项目(No.LY14E050024)
浙江省公益性技术应用研究项目(No.14609)
关键词
马尔可夫过程
转移概率矩阵
销售预测
Markov process
transition probability matrix
sales forecast
