移動(dòng)通信大數(shù)據(jù)分析：數(shù)據(jù)挖掘與機(jī)器學(xué)習(xí)實(shí)戰(zhàn)

第1章概述
1.1 電信業(yè)大數(shù)據(jù)分析 ···························1
1.2 電信大數(shù)據(jù)分析的驅(qū)動(dòng)力 ················2
1.3 大數(shù)據(jù)分析對(duì)電信產(chǎn)業(yè)價(jià)值鏈的
益處 ··················································3
1.4 電信大數(shù)據(jù)的實(shí)現(xiàn)范圍····················4
1.4.1 網(wǎng)絡(luò)分析 ···················································5
1.4.2 用戶與市場(chǎng)分析 ·······································8
1.4.3 創(chuàng)新的商業(yè)模式 ·······································91.5 本書(shū)概要 ··········································9
參考文獻(xiàn) ·················································10
第2章電信分析方法論
2.1 回歸方法 ········································12
2.1.1 線性回歸 ··················································13
2.1.2 非線性回歸 ··············································15
2.1.3 特征選擇 ··················································16
2.2 分類(lèi)方法 ········································18
2.2.1 邏輯回歸 ··················································18
2.2.2 其他分類(lèi)方法 ··········································19
2.3 聚類(lèi)方法 ········································20
2.3.1 K均值聚類(lèi) ··············································21
2.3.2 高斯混合模型 ··········································23
2.3.3 其他聚類(lèi)方法 ··········································24
2.3.4 聚類(lèi)方法在電信數(shù)據(jù)中的應(yīng)用 ·················25
2.4 預(yù)測(cè)方法 ········································25
2.4.1 時(shí)間序列分解 ··········································26
2.4.2 指數(shù)平滑模型 ··········································27
2.4.3 ARIMA模型 ············································28
2.5 神經(jīng)網(wǎng)絡(luò)和深度學(xué)習(xí) ·····················29
2.5.1 神經(jīng)網(wǎng)絡(luò) ··················································29
2.5.2 深度學(xué)習(xí) ··················································31
2.6 強(qiáng)化學(xué)習(xí) ········································32
2.6.1 模型和策略 ··············································33
2.6.2 強(qiáng)化學(xué)習(xí)算法 ··········································33
參考文獻(xiàn) ·················································34

XII
XII

第3章 LTE網(wǎng)絡(luò)性能趨勢(shì)分析
3.1 網(wǎng)絡(luò)性能預(yù)測(cè)策略 ·························39
3.1.1 直接預(yù)測(cè)策略 ··········································39
3.1.2 分析模型 ··················································39
3.2 網(wǎng)絡(luò)資源與性能指標(biāo)之間的關(guān)系 ···40
3.2.1 LTE網(wǎng)絡(luò)KPI與資源之間的關(guān)系 ···········40
3.2.2 回歸模型 ··················································41
3.3 網(wǎng)絡(luò)資源預(yù)測(cè) ·································43
3.3.1 LTE網(wǎng)絡(luò)流量與資源預(yù)測(cè)模型 ···············43
3.3.2 預(yù)測(cè)網(wǎng)絡(luò)資源 ··········································43
3.4 評(píng)估RRC連接建立的應(yīng)用 ············46
3.4.1 數(shù)據(jù)準(zhǔn)備與特征選取 ······························46
3.4.2 LTE KPI與網(wǎng)絡(luò)資源之間的關(guān)系推導(dǎo) ····47
3.4.3 預(yù)測(cè)RRC連接建立成功率 ·····················49
參考文獻(xiàn) ·················································50
第4章熱門(mén)設(shè)備就緒和返修率分析
4.1 設(shè)備返修率與設(shè)備就緒的預(yù)測(cè)
策略 ················································53
4.2 設(shè)備返修率和就緒預(yù)測(cè)模型 ··········54
4.2.1 預(yù)測(cè)模型的移動(dòng)通信服務(wù) ························54
4.2.2 參數(shù)獲取與存儲(chǔ) ······································55
4.2.3 分析引擎 ··················································56
4.3 實(shí)現(xiàn)和結(jié)果 ·····································58
4.3.1 設(shè)備返修率預(yù)測(cè) ······································58
4.3.2 設(shè)備就緒預(yù)測(cè) ··········································62
第5章 VoLTE語(yǔ)音質(zhì)量評(píng)估
5.1 應(yīng)用POLQA評(píng)估語(yǔ)音質(zhì)量··········68
5.1.1 POLQA標(biāo)準(zhǔn)···········································68
5.1.2 語(yǔ)音質(zhì)量評(píng)價(jià)中的可擴(kuò)展性和
可診斷性 ··················································69
5.2 CrowdMi方法論 ····························69
5.2.1 基于RF特征的分類(lèi) ·······························70
5.2.2 網(wǎng)絡(luò)指標(biāo)選擇與聚類(lèi) ······························70
5.2.3 網(wǎng)絡(luò)指標(biāo)與POLQA評(píng)分之間的關(guān)系····70
5.2.4 模型測(cè)試 ··················································70
5.3 CrowdMi中的技術(shù)細(xì)節(jié) ·················71
5.3.1 記錄分類(lèi) ··················································71
5.3.2 網(wǎng)絡(luò)指標(biāo)的選擇 ······································71
5.3.3 聚類(lèi) ·························································72
5.3.4 回歸 ·························································73
5.4 CrowdMi原型設(shè)計(jì)與試驗(yàn) ·············74
5.4.1 客戶端和服務(wù)器架構(gòu) ······························74
5.4.2 測(cè)試和結(jié)果 ··············································76
參考文獻(xiàn) ·················································78

目 錄XIII

目 錄XIII
第6章移動(dòng)APP無(wú)線資源使用分析
6.1 起因和系統(tǒng)概述 ·····························80
6.1.1 背景和挑戰(zhàn) ··············································80
6.1.2 移動(dòng)資源管理 ··········································81
6.1.3 系統(tǒng)概述 ··················································82
6.2 AppWiR眾包工具 ··························83
6.3 AppWiR挖掘算法 ··························84
6.3.1 網(wǎng)絡(luò)指標(biāo)的選擇 ······································84
6.3.2 LOESS方法 ············································87
6.3.3 基于時(shí)間序列的網(wǎng)絡(luò)資源使用預(yù)測(cè) ·······87
6.4 實(shí)現(xiàn)和試驗(yàn) ·····································88
6.4.1 數(shù)據(jù)收集與研究 ······································88
6.4.2 結(jié)果和準(zhǔn)確度 ··········································89
參考文獻(xiàn) ·················································91
第7章電信數(shù)據(jù)的異常檢測(cè)
7.1 模型 ················································93
7.1.1 高斯模型 ··················································94
7.1.2 時(shí)間依賴的高斯模型 ······························94
7.1.3 高斯混合模型（GMM）·························95
7.1.4 時(shí)間依賴的高斯混合模型 ·······················95
7.1.5 高斯概率潛在語(yǔ)義模型（GPLSA）·······95
7.2 模型對(duì)比 ········································97
7.2.1 樣本定義 ··················································97
7.2.2 異常識(shí)別 ··················································98
7.2.3 時(shí)間依賴GMM與GPLSA的對(duì)比 ·········997.3 仿真與討論 ···································100
參考文獻(xiàn) ···············································103
第8章基于大數(shù)據(jù)分析的LTE網(wǎng)絡(luò)自優(yōu)化
8.1 SON（自組織網(wǎng)絡(luò)）···················105
8.2 APP-SON ······································107
8.3 APP-SON架構(gòu) ·····························108
8.4 APP-SON算法 ·····························110
8.4.1 匈牙利算法輔助聚類(lèi)（HAAC）··········111
8.4.2 單位回歸輔助聚類(lèi)數(shù)的確定 ·················114
8.4.3 基于DNN的回歸·································114
8.4.4 每個(gè)小區(qū)在時(shí)序空間的標(biāo)簽組合 ·········116
8.4.5 基于相似性的參數(shù)調(diào)整 ·························1168.5 仿真與討論 ···································117
參考文獻(xiàn) ···············································122
第9章電信數(shù)據(jù)和市場(chǎng)營(yíng)銷(xiāo)
9.2.1 數(shù)據(jù)采集和數(shù)據(jù)類(lèi)型 ····························130
9.1 電信營(yíng)銷(xiāo)專(zhuān)題 ·······························127
9.2.2 網(wǎng)絡(luò)的提取和管理 ································131
9.2 社交網(wǎng)絡(luò)的總體構(gòu)建 ···················130



9.3 網(wǎng)絡(luò)結(jié)構(gòu)的度量 ···························133
參考文獻(xiàn) ···············································135

9.4 網(wǎng)絡(luò)中的消費(fèi)者行為建模 ············134
第10章傳染式客戶流失
10.1 問(wèn)題引入 ·····································138
10.1.1 流失率問(wèn)題 ··········································138
10.1.2 社交學(xué)習(xí)和網(wǎng)絡(luò)效應(yīng) ··························139
10.2 網(wǎng)絡(luò)數(shù)據(jù)的處理 ·························141
10.3 動(dòng)態(tài)模型 ·····································143
10.3.1 模型介紹 ··············································143
10.3.2 模型的定義 ··········································144
10.3.3 自身經(jīng)驗(yàn)建模、社交學(xué)習(xí)和
社交網(wǎng)絡(luò)效應(yīng) ······································146
10.3.4 模型估計(jì) ··············································148
10.4 結(jié)果 ············································149
參考文獻(xiàn) ···············································151
第11章基于社交網(wǎng)絡(luò)的精準(zhǔn)營(yíng)銷(xiāo)
11.1 網(wǎng)絡(luò)效應(yīng)的渠道 ·························158
11.2 社交網(wǎng)絡(luò)數(shù)據(jù)處理 ·····················159
11.3 建模策略問(wèn)題 ·····························160
11.3.1 線性空間自回歸模式 ···························160
11.3.2 社交網(wǎng)絡(luò)交互模型 ······························162
11.3.3 內(nèi)生同伴效應(yīng) ······································162
11.4 發(fā)現(xiàn)與應(yīng)用 ·································164
11.4.1 結(jié)果的解釋 ··········································164
11.4.2 基于社交網(wǎng)絡(luò)的精準(zhǔn)營(yíng)銷(xiāo) ···················165
參考文獻(xiàn) ···············································168
第12章社交影響和動(dòng)態(tài)社交網(wǎng)絡(luò)結(jié)構(gòu)
12.1 動(dòng)態(tài)模型 ·····································17712.1.1 連續(xù)時(shí)間馬爾可夫模型假設(shè) ···············17712.1.2 模型估計(jì)與識(shí)別 ··································17912.1.3 網(wǎng)絡(luò)結(jié)構(gòu)對(duì)社交影響的多元分析 ·······18012.2 研究發(fā)現(xiàn)總結(jié) ·····························18112.2.1 隨機(jī)行動(dòng)者動(dòng)態(tài)網(wǎng)絡(luò)模型的
估計(jì)結(jié)果··············································182

12.2.2 元回歸分析結(jié)果 ··································184
12.2.3 策略模擬 ··············································18812.3 結(jié)論 ············································193
參考文獻(xiàn) ···············································194