黑龙江烟草工业有限责任公司卷烟销量预测研究

发布时间：2018-03-17 05:28

本文选题：黑龙江　切入点：烟草企业　出处：《哈尔滨理工大学》2017年硕士论文　论文类型：学位论文

【摘要】：中国烟草业多年来全面推进“以客户订单为基准的组织货源”工作,这也正迎合了当今社会“以销定产”的思想,因此如何准确的预测销量,准确及时地反应烟草市场走势,为参与市场的生产者、营销方提供准确有效的决策参考依据,已成为市场各方最为急迫的要求。本文,首先介绍了黑龙江卷烟在国内的销售现状,分析了影响黑龙江省烟草有限责任公司(后文简称为龙江烟草公司)销量的具体因素,在模型的选择上介绍了灰色系统和马尔可夫模型的基本理论,说明了灰色预测模型和马尔可夫预测模型的概念和研究方法。在灰色系统理论基础上建立的灰色系统预测模型使用数据少,方便运算,预测准确度高。而马尔可夫链理论的研究适用于随机状态的动态系统,更加适合预测随机波动幅度比较大以及样本数据多的随机过程。但其不足之处是不仅要求预测对象具有马氏性,同时,预测数据序列要服从典型分布,但当系统信息较少时较难确定数据序列服从何种分布。所以,本文在分别使用灰色系统和马尔可夫模型对卷烟销量进行预测,得到结果分析各自模型的特点及优劣后,又将灰色预测模型和马尔可夫预测模型有机地结合起来,利用各自的模型优势,用灰色预测模型来预测卷烟销量的时间走势,随后对灰色产生的残差走势建立马尔可夫模型进行修正,从而使两者相结合建立更为精确的灰色马尔可夫预测模型,对各自模型取长补短,进行优化。本文对龙江烟草公司目前销售情况进行了研究,对2012-2015年共四年来每月的销量数据进行分析测算,使用灰色模型和马尔可夫模型及两个模型相结合建立的灰色马尔可夫模型,通过灰色模型及其相应的建模软件,对年销量样本数据进行分析、预测;对月销量样本数据应用马尔可夫模型结果进行对比,发现误差较大;最后通过将灰色模型的误差作为马尔可夫模型的区间划分依据,从而建立灰色马尔可夫模型,对季度的卷烟品类销量进行预测,取得了更为精确的预测结果。
[Abstract]:Over the years, the Chinese tobacco industry has comprehensively promoted the work of "organizing and supplying goods based on customer orders," which is also catering to the idea of "producing by sales" in today's society. Therefore, how to accurately predict the sales volume and accurately and timely reflect the trend of the tobacco market, For the producers participating in the market, the marketer provides the accurate and effective reference basis for decision-making, which has become the most urgent requirement of all parties in the market. This paper first introduces the current situation of Heilongjiang cigarette sales in China. This paper analyzes the specific factors influencing the sales volume of Heilongjiang Tobacco Co., Ltd. (hereinafter referred to as Longjiang Tobacco Company), and introduces the basic theories of grey system and Markov model in the selection of models. The concept and research method of grey prediction model and Markov prediction model are explained. The grey system prediction model established on the basis of grey system theory uses less data and is convenient for calculation. The Markov chain theory is suitable for dynamic systems with random states. It is more suitable to predict random processes with large amplitude of random fluctuation and large sample data. However, its shortcoming is that it not only requires the prediction object to have Markov property, but also the prediction data series should be distributed from a typical model. But when the system information is less, it is difficult to determine the distribution of the data sequence clothing. Therefore, the gray system and Markov model are used to predict the cigarette sales volume, and the results of the analysis of the characteristics, advantages and disadvantages of the respective models. Combining the grey forecasting model with the Markov forecasting model organically, using their respective model advantages, using the grey forecast model to predict the time trend of cigarette sales. Then, the Markov model is established to correct the residual trend of grey, so as to establish a more accurate grey Markov prediction model, which can make up for the weakness of each model. In this paper, the current sales situation of Longjiang Tobacco Company is studied, and the monthly sales data of 2012-2015 are analyzed and calculated. Using the grey model and Markov model and the grey Markov model established by the combination of the two models, through the grey model and its corresponding modeling software, the sample data of annual sales volume are analyzed and forecasted. By comparing the data of monthly sales volume with the results of Markov model, it is found that the error of grey model is large. Finally, the error of grey model is taken as the basis of interval division of Markov model, and then the grey Markov model is established. Forecast the quarterly cigarette sales and obtain more accurate forecast results.
【学位授予单位】：哈尔滨理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F426.8;F274

【参考文献】