组合预测方法在云产卷烟需求预测中的应用研究

发布时间：2018-07-31 07:53

【摘要】：自改革开放以来,中国烟草历经由小变大、由弱变强、由散乱到集中、由杂牌到大牌等发展历程。目前,受到国内外环境变化、经济下行压力的影响,中国烟草正处在销量增长难度加大、结构提升遭遇瓶颈、供需不平衡的、产业整体发展速度趋缓的形势,为此,本文根据行业发展趋势,根据“卷烟市场化取向改革”措施,以“需求预测”为落脚点,结合卷烟营销工作特点,对云产卷烟未来几年的需求进行预测,为云南中烟公司进行发展战略规划、目标决策制定和卷烟任务分配等发展方面提供有作用的信息资料和参考意见。首先,本文阐述了预测理论知识概念、类别和作用,描述了国内外预测方法的发展和使用情况,进一步比较了不同预测方法模型的优缺点,并着重分析了本文所采用的模型“组合预测模型”。其次,剖析了云产卷烟在国内卷烟市场的表现、竞争程度、品牌状态、一二类卷烟销售的态势等等；分析了行业发展形势和市场化取向改革推进,总结得出卷烟需求预测工作实施势在必行,将成为未来卷烟营销工作的核心和重点。接下来,通过重要性、经济性等多维度分析找出影响云产卷烟预测的八个要素。结合客观条件,收集采用新华社云南分社、国家烟草专卖局一号工程系统、中华人民共和国国家统计局以及其他参考资料中的数据,运用SPSS软件进行了逐步回归分析,将不符合检验的变量逐一剔除,得出与卷烟销售关系最紧密、关联度最高的三个要素,并使用“多元回归模型”对卷烟销售进行预测。由于该模型对卷烟政策因素、重大技改等客观影响反应不灵敏,而这些因素对卷烟销售极其重要,所以使得预测的结果带有一定的局限性。为此,本文又采用了“二次指数平滑法”,增加考虑周期性波动、环境变化等客观情况,再次对卷烟销售进行预测,并对两种不同预测方法的结果进行比较,发现“多元回归模型”对历史数据信息运用充分,但缺少对环境、政策等客观因素的考虑；而“二次指数平滑法”能充分考虑客观影响,但对数据信息使用又不如“多元回归模型”,综合两种方法的运用,就能互相弥补,规避缺陷,所以本文最终选用“组合预测模型”对卷烟销售进行了预测,分析发现此方法的预测结果确实与未来销售趋势最贴近,预测精度也比较高,对今后的卷烟销售、产品规划等有很重要的参考价值。最后,本文通过分析未来国家政策方向、行业发展环境,云南卷烟发展前景,可能面临的机遇和挑战,参照预测结果数据,对卷烟营销情况作出预判,在市场运作、卷烟价值提升、资源优化、科技创新四个方面对云产卷烟发展提出建议和意见。
[Abstract]:Since the reform and opening up, Chinese tobacco has experienced from small to large, from weak to strong, from scattered to centralized, from miscellaneous to big brand. At present, affected by the changes in the domestic and foreign environment and the downward pressure on the economy, Chinese tobacco is in a situation where the growth of sales volume is becoming more difficult, the structural upgrading is facing bottlenecks, the supply and demand are not balanced, and the overall development rate of the industry is slowing down. For this reason, According to the development trend of the industry, according to the measures of "cigarette market-oriented reform", taking "demand forecasting" as the foothold, combining with the characteristics of cigarette marketing, this paper forecasts the demand of cloud cigarettes in the next few years. It provides useful information and reference for the development of Yunnan Zhongyan Company in terms of development strategy planning, target decision making and cigarette task allocation. First of all, this paper describes the concepts, categories and functions of prediction theory, describes the development and use of prediction methods at home and abroad, and further compares the advantages and disadvantages of different prediction methods. The combined prediction model used in this paper is analyzed emphatically. Secondly, it analyzes the performance, competition degree, brand status, sales situation of one or two types of cigarettes in the domestic cigarette market, and analyzes the development situation of the industry and the promotion of market-oriented reform. It is concluded that the implementation of cigarette demand prediction is imperative and will become the core and focus of cigarette marketing in the future. Then, through the analysis of importance, economy and other multi-dimensional analysis, we find out the eight factors that affect cloud cigarette prediction. In combination with objective conditions, the data collected from the Yunnan Branch of Xinhua News Agency, the No.1 Project system of the State Tobacco Monopoly Administration, the National Bureau of Statistics of the people's Republic of China and other reference materials were collected, and the stepwise regression analysis was carried out using SPSS software. The variables of nonconformity test were eliminated one by one, and the three most closely related factors with cigarette sales were obtained, and the multiple regression model was used to predict cigarette sales. Because the model is not sensitive to the objective effects such as cigarette policy factors, major technical changes and so on, and these factors are extremely important to cigarette sales, the predicted results are limited to a certain extent. For this reason, this paper adopts "quadratic exponential smoothing" to increase the objective situation such as periodic fluctuation and environmental change, and forecasts the cigarette sales again, and compares the results of two different forecasting methods. It is found that "multivariate regression model" fully applies historical data information, but lacks consideration of objective factors such as environment and policy, while "quadratic exponential smoothing" can fully consider the objective influence. However, the use of data information is not as good as "multiple regression model", and the application of the two methods can make up for each other and avoid the defects, so this paper finally chooses "combination forecasting model" to forecast cigarette sales. It is found that the prediction results of this method are indeed the closest to the future sales trend and the prediction accuracy is also relatively high. It has important reference value for cigarette sales and product planning in the future. Finally, by analyzing the future national policy direction, the industry development environment, the development prospect of Yunnan cigarette, the opportunities and challenges that may be faced, and referring to the forecast result data, this paper makes a pre-judgment on the cigarette marketing situation and operates in the market. Four aspects of cigarette value promotion, resource optimization and scientific and technological innovation are put forward for the development of cloud cigarettes.
【学位授予单位】：昆明理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：F426.8;F273

【相似文献】