本文選題：運營商　切入點：流量經(jīng)營　出處：《南京郵電大學》2017年碩士論文　論文類型：學位論文

【摘要】：隨著移動互聯(lián)網(wǎng)的高速發(fā)展,運營商傳統(tǒng)的語音、短信業(yè)務收入明顯下降,各種移動互聯(lián)網(wǎng)業(yè)務快速發(fā)展,移動數(shù)據(jù)流量正急速增長。在流量急速增長的同時,運營商面臨著增量不增收的問題,而且有被互聯(lián)網(wǎng)公司管道化的威脅。在這樣的情況下,運營商做好流量經(jīng)營尤為重要。本文通過分析運營商流量經(jīng)營的必要性及優(yōu)缺點所在,通過建立數(shù)據(jù)挖掘模型,對用戶流量數(shù)據(jù)進行分析研究。以某運營商2015年6月份的南京用戶流量數(shù)據(jù)為基礎,采用遠程數(shù)據(jù)采集、本地環(huán)境搭建、本地分析用戶流量數(shù)據(jù)的思路,從客戶細分、上網(wǎng)偏好劃分、關聯(lián)規(guī)則尋找三個方面分析流量經(jīng)營中的問題:客戶細分主要以客戶流量行為、消費特征為指標,采取k-means算法對用戶進行細分,在結合業(yè)務知識的基礎上,最終細分為高、中、低三個檔次的用戶群;上網(wǎng)偏好劃分采取高頻偏好與普通偏好分離的方法,分別對其聚類,并分別得到4種單偏好群體、1種多偏好群體與9種單偏好群體、8種多偏好群體;然后加入用戶終端信息,通過FP-growth算法挖掘它們之間的強規(guī)則,得到16條終端、客戶檔次、上網(wǎng)偏好方面有價值的關聯(lián)規(guī)則,最終得出結論:新聞、視頻、財經(jīng)多偏好用戶更傾向發(fā)展為低檔次客戶;高檔次客戶更喜歡科技、閱讀和生活服務類的上網(wǎng)偏好;當用戶使用4G制式的手機時,使用華為品牌的置信度為65%,高于蘋果品牌的62%。本文的研究為運營商流量經(jīng)營提供了分析思路,完善系統(tǒng)及操作界面后,可作為運營商在流量經(jīng)營中進行針對性營銷的輔助工具。
[Abstract]:With the rapid development of mobile Internet, the revenue of traditional voice and short message service of operators has decreased obviously, various mobile Internet services have developed rapidly, and mobile data traffic is growing rapidly. At the same time of rapid growth of traffic, Operators are faced with the problem of no incremental revenue increase, and there is a threat of being managed by Internet companies. Under such circumstances, it is particularly important for operators to do traffic management well. This paper analyzes the necessity, advantages and disadvantages of operators' traffic management. Through the establishment of data mining model, the user traffic data are analyzed and studied. Based on the Nanjing user traffic data of a certain operator in June 2015, the remote data collection is adopted and the local environment is built. Local analysis of user traffic data ideas, from customer segmentation, Internet preference division, association rules looking for three aspects of traffic management problems: customer segmentation is mainly based on customer traffic behavior, consumption characteristics as indicators, The k-means algorithm is adopted to subdivide the users, and on the basis of combining business knowledge, it is finally subdivided into three user groups of high, middle and low levels; the division of Internet preference adopts the method of separating high frequency preference from common preference, and clustering them respectively. Four single preference groups, one multi-preference group and nine single-preference groups are obtained, and then the user terminal information is added, and the strong rules between them are mined by FP-growth algorithm, and 16 terminals and customer grades are obtained. The valuable association rules in the aspect of internet preference are concluded: news, video, finance and economics prefer users to develop into low-grade customers, high-grade customers prefer technology, reading and life service. When users use 4G mobile phones, the confidence of using Huawei brand is 65, which is higher than that of Apple brand. Can be used as operators in traffic management targeted marketing tools.
【學位授予單位】：南京郵電大學
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：F626;TP311.13

【參考文獻】

相關期刊論文 前7條

1 姚炯;;4G時代電信運營商流量經(jīng)營策略研究[J];現(xiàn)代營銷(下旬刊);2014年11期

2 蔣建洪;羅玫;;電子商務賣方數(shù)據(jù)聚類方法的改進研究[J];計算機工程與應用;2013年08期

3 薛京花;劉震宇;崔適時;;對K-means算法初始聚類中心選取的優(yōu)化[J];電子世界;2012年05期

4 談燕;賀仁龍;邱晨旭;;移動互聯(lián)網(wǎng)流量經(jīng)營數(shù)據(jù)分析與案例研究[J];電信科學;2012年02期

5 邱曉輝;;知識發(fā)現(xiàn)與數(shù)據(jù)挖掘分析[J];情報探索;2011年01期

6 謝宏,程浩忠,牛東曉;基于信息熵的粗糙集連續(xù)屬性離散化算法[J];計算機學報;2005年09期

7 陸楠,王U，

本文編號：1561179