基于对数线性模型的房地产广告费效分析

发布时间：2018-01-19 19:08

本文关键词： 对数线性模型 Monte Carlo模拟 VaR SAS 广告　出处：《重庆大学》2014年硕士论文　论文类型：学位论文

【摘要】：对数线性模型是广义线性模型中的一个分支，同Logistic模型、Probit模型一样，在定性数据的分析当中发挥着重要的作用。本文系统讲述了对数线性模型的基本原理，并且在此基础上也同时综合运用了VaR以及Monte Carlo模拟等方法，对房地产广告的类型和来访量之间的关系进行详细的分析，得出结论并给出合理的建议。通过研究分析表明，本文得出以下结论：（1）来访量会随着月份的变化而呈规律性的变化——先增后减，在二月份达到峰值，然后在九月份达到低谷；故建议在二月份左右适当的多投些广告量，在九月份左右适当的减少广告的投放；（2）从全年来评估效果来看，由大到小依次为：晨报、商报、时报、搜房网、短信、晚报、活动、日报、搜狐网、其他网媒；（3）来访量较小的时候，广告对来访量的影响比较大，，能够带来较大的增长；当来访量达到一定的程度时，广告的影响力就会大大降低；（4）在六个广告数量内随机投放，置信水平95%下的来访量与实际差距不是太大，因此，对于广告所能带来的来访量没有必要过分担心；（5）本文推荐选取晨报、晚报、商报、时报、短信、搜房网作为广告投放的重点，然后随机投放，在这些广告投放的过程中，只需考虑公司所能承受的广告成本即可。
[Abstract]:Logarithmic linear model is a branch of generalized linear model, just like Logistic model and probit model. It plays an important role in the analysis of qualitative data. This paper systematically describes the basic principle of logarithmic linear model. On the basis of this, we also use the methods of VaR and Monte Carlo simulation to analyze the relationship between the type of real estate advertisement and the quantity of visit in detail. Draw conclusions and give reasonable advice. The research results show that the following conclusions are drawn: 1) the number of visits will change regularly with the change of month, first increasing and then decreasing, reaching the peak on February, then reaching the low point on September; Therefore, it is suggested that the appropriate amount of advertising should be put in around February, and the appropriate amount of advertising should be reduced around September; (2) judging from the evaluation effect of the whole year, the order from big to small is: morning paper, business newspaper, times, search house net, short message, evening newspaper, activity, daily newspaper, Sohu net, other net medium; 3) when the number of visitors is small, the influence of advertising on the volume of visits is relatively large, which can bring about a greater growth; When the number of visitors reaches a certain level, the influence of advertising will be greatly reduced; 4) in the six advertisement quantity randomly, the confidence level 95% visit quantity and the actual disparity is not too big, therefore, for the advertisement can bring the visitor quantity does not need to worry too much; This article recommends choosing the morning newspaper, evening newspaper, business newspaper, times, short message, search house net as the focus of advertising, and then randomly put them in the process of these advertisements. Just consider the advertising costs the company can afford.
【学位授予单位】：重庆大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：O212;F299.233.42

【参考文献】