第1 章機(jī)器學(xué)習(xí) ··············································································································1
1.1 機(jī)器學(xué)習(xí)簡介·······································································································1
1.1.1 機(jī)器學(xué)習(xí)的概念······························································································1
1.1.2 機(jī)器學(xué)習(xí)的應(yīng)用領(lǐng)域························································································1
1.2 機(jī)器學(xué)習(xí)通用流程································································································2
1.2.1 目標(biāo)分析·······································································································2
1.2.2 數(shù)據(jù)準(zhǔn)備·······································································································3
1.2.3 特征工程·······································································································4
1.2.4 模型訓(xùn)練與調(diào)優(yōu)······························································································5
1.2.5 性能度量與模型應(yīng)用························································································6
1.3 Python 機(jī)器學(xué)習(xí)工具庫簡介·················································································6
1.3.1 數(shù)據(jù)準(zhǔn)備相關(guān)工具庫························································································6
1.3.2 數(shù)據(jù)可視化相關(guān)工具庫·····················································································7
1.3.3 模型訓(xùn)練與評估相關(guān)工具庫···············································································8
小結(jié)····························································································································9
課后習(xí)題 ··················································································································.10
第 2 章數(shù)據(jù)準(zhǔn)備 ···········································································································.12
2.1 數(shù)據(jù)質(zhì)量校驗····································································································.12
2.1.1 一致性校驗·································································································.12
2.1.2 缺失值校驗·································································································.15
2.1.3 異常值校驗·································································································.17
2.2 數(shù)據(jù)分布與趨勢探查·························································································.18
2.2.1 分布分析····································································································.18
2.2.2 對比分析····································································································.22
2.2.3 描述性統(tǒng)計分析···························································································.25
2.2.4 周期性分析·································································································.28
2.2.5 貢獻(xiàn)度分析·································································································.29
2.2.6 相關(guān)性分析·································································································.31
VIII
2.3 數(shù)據(jù)清洗···········································································································.35
2.3.1 缺失值處理·································································································.35
2.3.2 異常值處理·································································································.38
2.4 數(shù)據(jù)合并···········································································································.39
2.4.1 數(shù)據(jù)堆疊····································································································.39
2.4.2 主鍵合并····································································································.43
小結(jié)·························································································································.45
課后習(xí)題 ··················································································································.45
第 3 章特征工程 ···········································································································.48
3.1 特征變換···········································································································.48
3.1.1 標(biāo)準(zhǔn)化·······································································································.48
3.1.2 獨(dú)熱編碼····································································································.54
3.1.3 離散化·······································································································.55
3.2 特征選擇···········································································································.58
3.2.1 子集搜索與評價···························································································.58
3.2.2 過濾式選擇·································································································.59
3.2.3 包裹式選擇·································································································.59
3.2.4 嵌入式選擇與L1 范數(shù)正則化···········································································.60
3.2.5 稀疏表示與字典學(xué)習(xí)·····················································································.61
小結(jié)·························································································································.63
課后習(xí)題 ··················································································································.63
第 4 章有監(jiān)督學(xué)習(xí) ········································································································.66
4.1 有監(jiān)督學(xué)習(xí)簡介································································································.66
4.2 性能度量···········································································································.66
4.2.1 分類任務(wù)性能度量························································································.66
4.2.2 回歸任務(wù)性能度量························································································.68
4.3 線性模型···········································································································.69
4.3.1 線性模型簡介······························································································.69
4.3.2 線性回歸····································································································.69
4.3.3 邏輯回歸····································································································.72
4.4 k 近鄰分類········································································································.75
4.5 決策樹··············································································································.78
4.5.1 決策樹簡介·································································································.78
4.5.2 ID3 算法·····································································································.79
4.5.3 C4.5 算法····································································································.81
4.5.4 CART 算法··································································································.83
4.6 支持向量機(jī)·······································································································.86
4.6.1 支持向量機(jī)簡介···························································································.86
4.6.2 線性支持向量機(jī)···························································································.87
4.6.3 非線性支持向量機(jī)························································································.91
4.7 樸素貝葉斯·······································································································.94
4.8 神經(jīng)網(wǎng)絡(luò)···········································································································.98
4.8.1 神經(jīng)網(wǎng)絡(luò)介紹······························································································.98
4.8.2 BP 神經(jīng)網(wǎng)絡(luò)································································································.99
4.9 集成學(xué)習(xí)···········································································································104
4.9.1 Bagging ······································································································104
4.9.2 Boosting ·····································································································106
4.9.3 Stacking ······································································································115
小結(jié)·························································································································116
課后習(xí)題 ··················································································································116
第 5 章無監(jiān)督學(xué)習(xí) ········································································································118
5.1 無監(jiān)督學(xué)習(xí)簡介································································································118
5.2 降維··················································································································118
5.2.1 PCA ··········································································································118
5.2.2 核化線性降維······························································································121
5.3 聚類任務(wù)···········································································································123
5.3.1 聚類性能度量指標(biāo)························································································124
5.3.2 距離計算····································································································125
5.3.3 原型聚類····································································································126
5.3.4 密度聚類····································································································137
5.3.5 層次聚類····································································································139
小結(jié)·························································································································142
課后習(xí)題 ··················································································································142
第 6 章智能推薦 ···········································································································144
6.1 智能推薦簡介····································································································144
6.1.1 推薦系統(tǒng)····································································································144
6.1.2 智能推薦的應(yīng)用···························································································144
6.2 推薦系統(tǒng)性能度量·····························································································146
6.2.1 離線實驗評價指標(biāo)························································································146
6.2.2 用戶調(diào)查評價指標(biāo)························································································148
6.2.3 在線實驗評價指標(biāo)························································································149
6.3 基于關(guān)聯(lián)規(guī)則的推薦技術(shù)··················································································149
6.3.1 關(guān)聯(lián)規(guī)則和頻繁項集·····················································································150
6.3.2 Apriori 算法·································································································150
6.3.3 FP-Growth 算法····························································································154
6.4 基于協(xié)同過濾的推薦技術(shù)··················································································159
6.4.1 基于用戶的協(xié)同過濾·····················································································159
6.4.2 基于物品的協(xié)同過濾·····················································································163
小結(jié)·························································································································166
課后習(xí)題 ··················································································································167
第 7 章醫(yī)療保險的欺詐發(fā)現(xiàn) ··························································································169
7.1 目標(biāo)分析···········································································································169
7.1.1 背景··········································································································169
7.1.2 數(shù)據(jù)說明····································································································170
7.1.3 分析目標(biāo)····································································································171
7.2 數(shù)據(jù)準(zhǔn)備···········································································································172
7.2.1 描述性統(tǒng)計分析···························································································172
7.2.2 數(shù)據(jù)清洗····································································································172
7.2.3 分析投保人和醫(yī)療機(jī)構(gòu)的信息·········································································173
7.3 特征工程···········································································································177
7.3.1 特征選擇····································································································177
7.3.2 特征變換····································································································178
7.4 模型訓(xùn)練···········································································································182
7.5 性能度量···········································································································184
7.5.1 結(jié)果分析····································································································184
7.5.2 聚類性能度量······························································································188
小結(jié)·························································································································190
第 8 章中醫(yī)證型關(guān)聯(lián)規(guī)則分析 ······················································································191
8.1 目標(biāo)分析···········································································································191
8.1.1 背景··········································································································191
8.1.2 數(shù)據(jù)說明····································································································191
8.1.3 分析目標(biāo)····································································································192
8.2 數(shù)據(jù)準(zhǔn)備···········································································································193
8.2.1 數(shù)據(jù)獲取····································································································193
8.2.2 數(shù)據(jù)清洗····································································································195
8.3 特征工程···········································································································196
8.3.1 特征選擇····································································································196
8.3.2 特征變換····································································································197
8.4 模型訓(xùn)練···········································································································201
8.5 性能度量···········································································································202
8.5.1 結(jié)果分析····································································································203
8.5.2 模型應(yīng)用····································································································204
小結(jié)·························································································································204
第 9 章糖尿病遺傳風(fēng)險預(yù)測 ··························································································205
9.1 目標(biāo)分析···········································································································205
9.1.1 背景··········································································································205
9.1.2 數(shù)據(jù)說明····································································································206
9.1.3 分析目標(biāo)····································································································207
9.2 數(shù)據(jù)準(zhǔn)備···········································································································207
9.2.1 數(shù)據(jù)探索····································································································207
9.2.2 數(shù)據(jù)清洗····································································································209
9.3 特征工程···········································································································209
9.4 模型構(gòu)建···········································································································211
9.4.1 交叉驗證····································································································211
9.4.2 模型訓(xùn)練····································································································213
9.5 性能度量···········································································································214
9.5.1 結(jié)果分析····································································································214
9.5.2 模型評價····································································································216
小結(jié)·························································································································216
第 10 章基于深度殘差神經(jīng)網(wǎng)絡(luò)的皮膚癌檢測································································217
10.1 目標(biāo)分析·········································································································217
10.1.1 背景·········································································································217
10.1.2 圖像數(shù)據(jù)說明·····························································································218
10.1.3 分析方法與過程··························································································219
10.2 圖像數(shù)據(jù)預(yù)處理······························································································219
10.2.1 圖像預(yù)處理································································································219
10.2.2 查看處理后的圖像·······················································································222
10.3 模型構(gòu)建·········································································································223
10.3.1 卷積神經(jīng)網(wǎng)絡(luò)（CNN） ················································································223
10.3.2 殘差網(wǎng)絡(luò)（Residual Network） ·······································································226
10.3.3 ImageDataGenerator 參數(shù)說明·········································································228
10.3.4 訓(xùn)練深度殘差神經(jīng)網(wǎng)絡(luò)模型···········································································229
10.4 性能度量·········································································································231
10.4.1 性能分析···································································································231
10.4.2 結(jié)果分析···································································································232
小結(jié)·························································································································234
第 11 章基于 TipDM 數(shù)據(jù)挖掘建模平臺實現(xiàn)醫(yī)療保險的欺詐發(fā)現(xiàn)··································236
11.1 TipDM 數(shù)據(jù)挖掘建模平臺················································································236
11.1.1 首頁·········································································································237
11.1.2 數(shù)據(jù)源······································································································238
11.1.3 工程·········································································································239
11.1.4 系統(tǒng)組件···································································································240
11.1.5 TipDM 數(shù)據(jù)挖掘建模平臺的本地化部署···························································241
11.2 快速構(gòu)建醫(yī)療保險的欺詐發(fā)現(xiàn)工程··································································243
11.2.1 獲取數(shù)據(jù)···································································································244
11.2.2 數(shù)據(jù)準(zhǔn)備···································································································247
11.2.3 特征工程···································································································250
11.2.4 模型訓(xùn)練···································································································253
小結(jié)·························································································································255
參考文獻(xiàn) ·························································································································256