基于改进KNN算法的二手房评估

发布时间：2018-07-18 11:03

【摘要】：传统的房屋评估方法,比如收益法、成本法、市场比较法等方法多存在成本高、效率低、精度差等各项问题。本论文基于对KNN算法和二手房评估的研究,分析了 KNN算法的特点以及将算法用于二手房评估上的可行性。二手房信息中的关键数据容易数值化和标准化,使用KNN算法在模型上可行;经过筛选后的样本集规模可控,KNN算法在时间复杂度上优化空间也很大,在计算效率上可行。对于二手房信息这类结构较清晰的数据,KNN算法较易于实现,成本较低,在经济上可行。本论文对数据挖掘中的分类技术和回归技术进行分析,之后选取KNN算法为对二手房进行评估的核心技术,实现了一个B/S(Browser/Server,浏览器/服务器模式)估价应用,给予有二手房评估需求的目标用户一种快速获得评估结果的方式。经研究分析,经典KNN算法具有精度较高、对样本集中的噪声不敏感等优点,也有k值难以选取、时间复杂度高、受样本平衡度的影响大等缺点。本着扬长避短的思想,对于算法优点,使用结果集加权的方式进一步提高算法精度,使用去重和标准化等方式减少噪声;对于算法缺点,使用多次检验法选定k值,使用TopK算法以及多线程并发的方式降低时间复杂度,使用数据采集阶段就对数据进行分类的方式稳定样本平衡度。为了对改进的KNN算法进行实用性验证,利用该算法对哈尔滨市的部分二手房数据进行分析,通过对数据的预处理,基于改进KNN算法的实现等步骤,最终给出二手房的估价结果。对于二手房业主及中介商等目标用户,使用改进KNN算法进行二手房评估的B/S应用对比传统的二手房评估方式,具有计算速度快、交互界面友好易用的优势,很好的满足了目标用户的需求。
[Abstract]:Traditional methods of house evaluation, such as income method, cost method and market comparison method, have many problems, such as high cost, low efficiency, poor precision and so on. Based on the research of KNN algorithm and second-hand house evaluation, this paper analyzes the characteristics of KNN algorithm and the feasibility of applying the algorithm to second-hand housing evaluation. The key data in second-hand housing information is easy to be numerical and standardized, and the KNN algorithm is feasible in the model, and the filtered sample set size controllable KNN algorithm also has a large space in time complexity optimization, and is feasible in computing efficiency. The KNN algorithm is easy to realize for the second-hand housing information, which has a clear structure, and the cost is lower, so it is economically feasible. This paper analyzes the classification technology and regression technology in data mining, then selects KNN algorithm as the core technology to evaluate second-hand housing, and realizes a B / S (browser / Server mode) evaluation application. Give target users with second-hand housing assessment needs a quick way to get results. Through research and analysis, it is found that the classical KNN algorithm has the advantages of high precision, insensitivity to the noise in the sample set, and the disadvantages of hard to select k value, high time complexity and large influence of sample balance. For the advantages of the algorithm, the method of weighted result set is used to further improve the accuracy of the algorithm, and the method of de-duplication and standardization is used to reduce the noise, and for the shortcomings of the algorithm, the value of k is selected by the method of multiple tests. The time complexity is reduced by using TopK algorithm and multi-thread concurrency, and the sample balance is stabilized by classifying the data in the data acquisition stage. In order to verify the practicability of the improved KNN algorithm, this algorithm is used to analyze some second-hand housing data in Harbin. Through the preprocessing of the data and the realization of the improved KNN algorithm, the evaluation results of the second-hand house are given. For the target users such as second-hand house owners and intermediaries, the B / S application of the improved KNN algorithm for second-hand housing evaluation has the advantages of faster calculation speed and friendly interface than the traditional second-hand housing evaluation method. Meet the needs of the target users well.
【学位授予单位】：哈尔滨商业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F299.23;TP311.13

【参考文献】