Building Machine Learning Systems with Python - Richert, Coelho

Table of Contents
Tuning the instance 90
Tuning the classifier 90
Fetching the data 91
Slimming the data down to chewable chunks 92
Preselection and processing of attributes 93
Defining what is a good answer 94
Creating our first classifier 95
Starting with the k-nearest neighbor (kNN) algorithm 95
Engineering the features 96
Training the classifier 97
Measuring the classifier's performance 97
Designing more features 98
Deciding how to improve 101
Bias-variance and its trade-off 102
Fixing high bias 102
Fixing high variance 103
High bias or low bias 103
Using logistic regression 105
A bit of math with a small example 106
Applying logistic regression to our postclassification problem 108
Looking behind accuracy – precision and recall 110
Slimming the classifier 114
Ship it! 115
Summary 115
Chapter 6: Classification II – Sentiment Analysis 117
Sketching our roadmap 117
Fetching the Twitter data 118
Introducing the Naive Bayes classifier 118
Getting to know the Bayes theorem 119
Being naive 120
Using Naive Bayes to classify 121
Accounting for unseen words and other oddities 124
Accounting for arithmetic underflows 125
Creating our first classifier and tuning it 127
Solving an easy problem first 128
Using all the classes 130
Tuning the classifier's parameters 132
Cleaning tweets 136
Taking the word types into account 138
Determining the word types 139
[ iii ]