Author(s): Jason Brownlee
Year: 0
Language: English
Pages: 115
Copyright......Page 1
I Introduction......Page 7
Book Organization......Page 8
Your Outcomes From Reading This Book......Page 10
Summary......Page 11
II XGBoost Basics......Page 13
Origin of Boosting......Page 14
Generalization of AdaBoost as Gradient Boosting......Page 15
How Gradient Boosting Works......Page 16
Improvements to Basic Gradient Boosting......Page 17
Summary......Page 19
What is XGBoost?......Page 20
XGBoost Features......Page 21
Why Use XGBoost?......Page 22
Summary......Page 23
Install XGBoost for Use in Python......Page 24
Load and Prepare Data......Page 25
Train the XGBoost Model......Page 26
Tie it All Together......Page 27
Summary......Page 28
Label Encode String Class Values......Page 29
One Hot Encode Categorical Data......Page 31
Support for Missing Data......Page 34
Summary......Page 37
Evaluate Models With Train and Test Sets......Page 38
Evaluate Models With k-Fold Cross-Validation......Page 39
Summary......Page 41
Plot a Single XGBoost Decision Tree......Page 42
Summary......Page 44
III XGBoost Advanced......Page 45
Serialize Models with Pickle......Page 46
Serialize Models with Joblib......Page 47
Summary......Page 49
Feature Importance in Gradient Boosting......Page 50
Manually Plot Feature Importance......Page 51
Using theBuilt-in XGBoost Feature Importance Plot......Page 52
Feature Selection with XGBoost Feature Importance Scores......Page 54
Summary......Page 55
Monitoring Training Performance With XGBoost......Page 56
Evaluate XGBoost Models With Learning Curves......Page 58
Early Stopping With XGBoost......Page 61
Summary......Page 62
Problem Description: Otto Dataset......Page 64
Impact of the Number of Threads......Page 65
Parallelism When Cross Validating XGBoost Models......Page 68
Summary......Page 70
Tutorial Overview......Page 71
Launch Your Server Instance......Page 72
Login and Configure......Page 75
Train an XGBoost Model......Page 78
Close Your AWS Instance......Page 79
Summary......Page 80
IV XGBoost Tuning......Page 81
Configuration Advice from Primary Sources......Page 82
Configuration Advice From scikit-learn......Page 84
Configuration Advice From XGBoost......Page 85
Summary......Page 87
Tune the Number of Decision Trees......Page 88
Tune the Size of Decision Trees......Page 91
Tune The Number and Size of Trees......Page 93
Summary......Page 96
Slow Learning in Gradient Boosting with a Learning Rate......Page 97
Tuning Learning Rate......Page 98
Tuning Learning Rate and the Number of Trees......Page 100
Summary......Page 103
Stochastic Gradient Boosting......Page 104
Tuning Row Subsampling......Page 105
Tuning Column Subsampling By Tree......Page 107
Tuning Column Subsampling By Split......Page 109
Summary......Page 111
V Conclusions......Page 112
How Far You Have Come......Page 113
Gradient Boosting in Textbooks......Page 114
XGBoost Library......Page 115
