基于大数据的住房租金预测模型

随着城市化进程的推进和社会经济的飞速发展，我国的房屋价格不断上涨，对于不少人来说，购买房屋有着一定的困难，所以在购房压力下，出现了一大批倾向于以租住房屋为主要生活方式的群体。然而，仍有一些棘手的问题存在于租住房屋领域，其中最普遍的矛盾——房东和租客信息不对称：其一，一般来讲房东将住房委托给房产中介，并不实际了解租房市场，不能对出租房屋的租金进行合理准确的评估；其二，租客通过租房中介得到的信息是经过修饰美化的，因此要找到与自身租房预期完美符合且性价比较高的住房并不是一件容易的事，信息不对称的问题产生了许多不良影响，不仅会浪费租房资源，还会使租赁市场发生波动，打破其稳定性。因此，本文建立住房租金的预测模型，对模型进行研究，旨在通过构建模型后得出的预测效果深入分析现存租赁市场的住房租金形式，从而更好地把握住房租赁市场价格变动的脉搏，使各方面问题迎刃而解。本文基于近二十万个样本，进行数据处理并建立多元线性回归、神经网络、随机森林模型，对比其预测准确性，并选择最优模型。

关键词：大数据，预测模型，模型比较

 
Abstract
With the advancement of urbanization and the rapid development of social economy, the price of houses in our country is constantly rising. For many people, it is difficult to buy houses. Therefore, under the pressure of buying houses, a large number of people tend to rent houses. Housing is the main lifestyle group. However, there are still some thorny problems in the field of rental housing. The most common contradiction is information asymmetry between landlords and tenants: First, landlords generally entrust housing to real estate agencies and do not actually know the rental market, so it is impossible to make a reasonable and accurate assessment of the rent of the rental housing. Second, the information obtained by the tenant through the rental agency is modified and beautified, so it is not easy to find a housing that perfectly matches their rental expectations and is more cost-effective The problem of information asymmetry has produced many undesirable effects, which not only wastes rental housing resources, but also causes fluctuations in the rental market and breaks its stability. Therefore, this article establishes a housing rent prediction model and conducts research on the model. The aim is to analyze the existing rental market’s housing rent form through the prediction effect obtained after constructing the model, so as to better grasp the pulse of housing rental market price changes and make All aspects of the problem can be solved easily. Based on nearly 200,000 samples, this paper conducts data processing and establishes multiple linear regression, neural network, and random forest models, compares their prediction accuracy, and selects the best model.
Keywords: big data, predictive models, model comparison
 
目录
1 绪论 7
1.1 研究目的及意义 7
1.2 国内外研究现状 7
1.2.1 国外研究现状 7
1.2.2 国内研究现状 8
1.3 文献综述总结 9
1.3.1 预测模型总结 9
1.3.2 发展情况总结 10
2 研究内容、方法与分析框架 11
2.1 研究内容 11
2.2 研究方法 11
2.3 技术路线图 12
3 数据获取与处理 13
3.1 数据基本情况 13
3.2 数据预处理 14
3.2.1 变量选择 14
3.2.2 剔除重复数据 14
3.2.3 删除无效变量 15
3.2.4 补充缺失值 15
3.2.5 替换异常值 20
3.3 变量分析 16
3.3.1 描述统计 16
3.3.2 相关性分析 18
3.2.3 数据标准化 20
3.4 划分训练集与测试集 20
4 建立模型 21
4.1 线性回归 21
4.2 神经网络 22
4.3 随机森林 22
4.3.1随机森林算法概述 22
4.3.2 随机森林算法特征 23
4.4 模型总结 23
5 模型比较及结论 24
6 存在问题及发展趋势 26
6.1 存在问题 26
6.2 发展趋势 26
参考文献 27

*若需了解更多与协助请咨询↓→[电脑QQ][手机QQ]【数据协助】