【摘要】：隨著數(shù)據(jù)庫技術(shù)的廣泛應(yīng)用,各行業(yè)積累了海量的業(yè)務(wù)數(shù)據(jù)。為了改變當(dāng)前“數(shù)據(jù)豐富卻知識匱乏”的狀況,數(shù)據(jù)挖掘技術(shù)得到了眾多企業(yè)的重視。在現(xiàn)今移動數(shù)據(jù)時代、智能終端時代下,智能手機(jī)越來越普遍,用戶換機(jī)的頻率和對手機(jī)的依賴性更加強(qiáng)烈。所以,在手機(jī)終端精準(zhǔn)營銷方向,對更換手機(jī)的潛在客戶預(yù)測這項研究是有意義的。 在這背景下,本文提出基于數(shù)據(jù)挖掘技術(shù)對換機(jī)客戶進(jìn)行預(yù)測研究的方法。首先,本文依托X省公司“手機(jī)終端精準(zhǔn)營銷需求”項目,分析和采集用戶行為數(shù)據(jù),并借助Hadoop對真實用戶上網(wǎng)日志進(jìn)行分析。參照CRISP-DM挖掘流程對采集的數(shù)據(jù)進(jìn)行數(shù)據(jù)理解、數(shù)據(jù)清洗、數(shù)據(jù)轉(zhuǎn)換等工作。其次,本文分別采用決策樹C5.0、神經(jīng)網(wǎng)絡(luò)、Logistic回歸算法對樣本數(shù)據(jù)進(jìn)行訓(xùn)練并建立預(yù)測模型,并對訓(xùn)練結(jié)果進(jìn)行評估和比較。通過SPSS的實驗結(jié)果表明決策樹C5.0算法模型對潛在換機(jī)預(yù)測研究更為適合。最后,本文分析了模型對于用戶市場拓展、開展終端精準(zhǔn)營銷、提升業(yè)務(wù)推薦成功率、用戶終端行為監(jiān)控等方面簡單的應(yīng)用情況。 本文從實際問題出發(fā),將數(shù)據(jù)挖掘技術(shù)應(yīng)用到潛在手機(jī)終端更換的預(yù)測研究中,研究工作對決策及市場人員開展工作有重要的作用,對同類型換機(jī)預(yù)測研究有一定參考意義。但本次研究還存在可以改進(jìn)的地方如評價指標(biāo)、業(yè)務(wù)知識,數(shù)據(jù)處理方法,這些也是將來可開展的研究工作。
[Abstract]:With the wide application of database technology, various industries have accumulated massive business data. In order to change the current situation of "data rich but lack of knowledge", data mining technology has been paid attention to by many enterprises. In the era of mobile data and intelligent terminal, smartphone is becoming more and more popular. So in the precise marketing direction of mobile terminals, it makes sense to predict the potential customers of mobile phone replacement. Under this background, this paper puts forward a method of forecasting customers based on data mining technology. First of all, this paper relies on X province company "the mobile phone terminal precision marketing demand" project, analyzes and collects the user behavior data, and carries on the analysis to the real user online log with the aid of Hadoop. According to the CRISP-DM mining process, data understanding, data cleaning, data conversion and so on. Secondly, the decision tree C5.0, neural network and Logistic regression algorithm are used to train the sample data and establish the prediction model, and the training results are evaluated and compared. The experimental results of SPSS show that the decision tree C5.0 algorithm model is more suitable for the prediction of potential machine change. Finally, this paper analyzes the simple application of the model for user market expansion, terminal precision marketing, promotion of business recommendation success rate, user terminal behavior monitoring and other aspects. Based on the practical problems, this paper applies the data mining technology to the prediction research of the potential mobile phone terminal replacement. The research work plays an important role in the decision making and the work of the market personnel, and has certain reference significance for the prediction research of the same type of machine exchange. However, there are still some areas that can be improved in this study, such as evaluation index, business knowledge, data processing methods, which are also the research work that can be carried out in the future.
【學(xué)位授予單位】：云南大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2015
【分類號】：TP311.13

【參考文獻(xiàn)】

相關(guān)期刊論文 前4條

1 仇春芳;李衛(wèi)衛(wèi);;數(shù)據(jù)挖掘在電信客戶流失分析中的應(yīng)用[J];通信世界;2007年05期

2 李仲俠;溫鋒;;把握數(shù)字家庭終端發(fā)展趨勢 運(yùn)營商踐行“定制終端”路線[J];通信世界;2012年09期

3 魯義軒;;管道建設(shè)融入流量經(jīng)營 智能化非惟一目標(biāo)[J];通信世界;2012年32期

4 郭明;鄭惠莉;;電信業(yè)呼吁數(shù)據(jù)挖掘[J];中國電信業(yè);2004年12期

，

本文編號：2399923