Python Machine Learning By Example电子书

Title Page

Copyright

Python Machine Learning By Example

Credits

About the Author

About the Reviewer

www.PacktPub.com

Customer Feedback

Preface

What this book covers

What you need for this book

Who this book is for

Conventions

Reader feedback

Customer support

Downloading the example code

Errata

Piracy

Questions

Getting Started with Python and Machine Learning

What is machine learning and why do we need it?

A very high level overview of machine learning

A brief history of the development of machine learning algorithms

Generalizing with data

Overfitting, underfitting and the bias-variance tradeoff

Avoid overfitting with cross-validation

Avoid overfitting with regularization

Avoid overfitting with feature selection and dimensionality reduction

Preprocessing, exploration, and feature engineering

Missing values

Label encoding

One-hot-encoding

Scaling

Polynomial features

Power transformations

Binning

Combining models

Bagging

Boosting

Stacking

Blending

Voting and averaging

Installing software and setting up

Troubleshooting and asking for help

Summary

Exploring the 20 Newsgroups Dataset with Text Analysis Algorithms

What is NLP?

Touring powerful NLP libraries in Python

The newsgroups data

Getting the data

Thinking about features

Visualization

Data preprocessing

Clustering

Topic modeling

Summary

Spam Email Detection with Naive Bayes

Getting started with classification

Types of classification

Applications of text classification

Exploring naive Bayes

Bayes' theorem by examples

The mechanics of naive Bayes

The naive Bayes implementations

Classifier performance evaluation

Model tuning and cross-validation

Summary

News Topic Classification with Support Vector Machine

Recap and inverse document frequency

Support vector machine

The mechanics of SVM

Scenario 1 - identifying the separating hyperplane

Scenario 2 - determining the optimal hyperplane

Scenario 3 - handling outliers

The implementations of SVM

Scenario 4 - dealing with more than two classes

The kernels of SVM

Choosing between the linear and RBF kernel

News topic classification with support vector machine

More examples - fetal state classification on cardiotocography with SVM

Summary

Click-Through Prediction with Tree-Based Algorithms

Brief overview of advertising click-through prediction

Getting started with two types of data, numerical and categorical

Decision tree classifier

The construction of a decision tree

The metrics to measure a split

The implementations of decision tree

Click-through prediction with decision tree

Random forest - feature bagging of decision tree

Summary

Click-Through Prediction with Logistic Regression

One-hot encoding - converting categorical features to numerical

Logistic regression classifier

Getting started with the logistic function

The mechanics of logistic regression

Training a logistic regression model via gradient descent

Click-through prediction with logistic regression by gradient descent

Training a logistic regression model via stochastic gradient descent

Training a logistic regression model with regularization

Training on large-scale datasets with online learning

Handling multiclass classification

Feature selection via random forest

Summary

Stock Price Prediction with Regression Algorithms

Brief overview of the stock market and stock price

What is regression?

Predicting stock price with regression algorithms

Feature engineering

Data acquisition and feature generation

Linear regression

Decision tree regression

Support vector regression

Regression performance evaluation

Stock price prediction with regression algorithms

Summary

Best Practices

Machine learning workflow

Best practices in the data preparation stage

Best practice 1 - completely understand the project goal

Best practice 2 - collect all fields that are relevant

Best practice 3 - maintain consistency of field values

Best practice 4 - deal with missing data

Best practices in the training sets generation stage

Best practice 5 - determine categorical features with numerical values

Best practice 6 - decide on whether or not to encode categorical features

Best practice 7 - decide on whether or not to select features and if so, how

Best practice 8 - decide on whether or not to reduce dimensionality and if so how

Best practice 9 - decide on whether or not to scale features

Best practice 10 - perform feature engineering with domain expertise

Best practice 11 - perform feature engineering without domain expertise

Best practice 12 - document how each feature is generated

Best practices in the model training, evaluation, and selection stage

Best practice 13 - choose the right algorithm(s) to start with

Naive Bayes

Logistic regression

SVM

Random forest (or decision tree)

Neural networks

Best practice 14 - reduce overfitting

Best practice 15 - diagnose overfitting and underfitting

Best practices in the deployment and monitoring stage

Best practice 16 - save, load, and reuse models

Best practice 17 - monitor model performance

Best practice 18 - update models regularly

Summary