What do you think?
Rate this book
Handbook of Anomaly Detection: With Python Outlier Detection: Build and modernize your anomaly detection models with examples
PyOD has been successfully used in numerous academic types of research and commercial products with more than 10 million downloads since 2017
“Anomaly detection is a critical technique to identify rare items in risk modeling, security, and healthcare. Dr. Kuo's book provides hands-on guidance on how to use existing tools, such as PyOD, to leverage this technique in your daily data science work. More importantly, this book reviews more than 10 leading detection algorithms, with both algorithm descriptions and detailed code examples.” ~ Yue Zhao, Principal Developer of PyOD
Who this book Is for
This book is for readers who are searching for anomaly detection methods or want to know about any new development in this field. Data science professionals in fields such as business, finance, insurance, engineering, science, or medical, will gain a deep understanding of the practices of different modeling techniques.
What you will learn
This book teaches readers how to think about anomalies and how to detect them successfully. The selected algorithms, from simple to complex, help readers expand to other advanced algorithms.
Ch. 1 presents real-world applications of anomaly detection, the characteristics of anomalies, and the general modeling strategies.
Ch. 2 and 3 present intuitive and fast algorithms including the Histogram-Based Outlier Score (HBOS), the Empirical Cumulative Distribution-based Outlier Detection (ECOD)
Ch. 4 explains how IForest is different from other types of models, and how effective it is in detecting outliers.
Ch. 5 describes how Principal Component Analysis (PCA) is applied in anomaly detection. You shall get a renewed perspective toward PCA.
Ch. 6 tells you how One-Class Support Vector Machine (OCSVM) emerges from SVM, and how it was invented for the use cases when anomalies are extremely hard to find.
Ch. 7 teaches you to detect anomalies through Gaussian Mixture Model (GMM). It illustrates the connection between GMM and K-means. It explains how Expectation-Maximization applies the Bayesian rule to find optimal parameters.
Ch. 8 helps you to build K-nearest Neighbors (KNN) models for anomaly detection.
Ch. 9 explains the distinction between global and local outliers. It illustrates how the Local Outlier Factor (LOF) can effectively detect local outliers. Chapter 10 extends to Cluster-Based Local Outlier Factor (CBLOF) to detect outliers.
Ch. 11 introduces the Extreme Gradient Boosting Outlier Detection (XGBOD). It presents a new concept called representation learning that studies systematic ways to discover representations for raw data without any human intervention. You will learn the outputs of unsupervised learning algorithms are fed to a supervised learning model to achieve model predictability.
Ch. 12 explains autoencoders and overall deep learning structure. You will apply autoencoders to detect anomalies. This chapter does not require readers to have deep learning background and teaches deep learning with a regression-friend approach.
The rarity of outliers implies the target will be extremely imbalanced in supervised learning models. Ch. 13 and 14 teach ten under- and over-sampling techniques to overcome the challenges.
Code examples
All code examples are in the GitHub folder. There is no need to type in the code.
“Anomaly detection is a critical technique to identify rare items in risk modeling, security, and healthcare. Dr. Kuo's book provides hands-on guidance on how to use existing tools, such as PyOD, to leverage this technique in your daily data science work. More importantly, this book reviews more than 10 leading detection algorithms, with both algorithm descriptions and detailed code examples.” ~ Yue Zhao, Principal Developer of PyOD
Who this book Is for
This book is for readers who are searching for anomaly detection methods or want to know about any new development in this field. Data science professionals in fields such as business, finance, insurance, engineering, science, or medical, will gain a deep understanding of the practices of different modeling techniques.
What you will learn
This book teaches readers how to think about anomalies and how to detect them successfully. The selected algorithms, from simple to complex, help readers expand to other advanced algorithms.
Ch. 1 presents real-world applications of anomaly detection, the characteristics of anomalies, and the general modeling strategies.
Ch. 2 and 3 present intuitive and fast algorithms including the Histogram-Based Outlier Score (HBOS), the Empirical Cumulative Distribution-based Outlier Detection (ECOD)
Ch. 4 explains how IForest is different from other types of models, and how effective it is in detecting outliers.
Ch. 5 describes how Principal Component Analysis (PCA) is applied in anomaly detection. You shall get a renewed perspective toward PCA.
Ch. 6 tells you how One-Class Support Vector Machine (OCSVM) emerges from SVM, and how it was invented for the use cases when anomalies are extremely hard to find.
Ch. 7 teaches you to detect anomalies through Gaussian Mixture Model (GMM). It illustrates the connection between GMM and K-means. It explains how Expectation-Maximization applies the Bayesian rule to find optimal parameters.
Ch. 8 helps you to build K-nearest Neighbors (KNN) models for anomaly detection.
Ch. 9 explains the distinction between global and local outliers. It illustrates how the Local Outlier Factor (LOF) can effectively detect local outliers. Chapter 10 extends to Cluster-Based Local Outlier Factor (CBLOF) to detect outliers.
Ch. 11 introduces the Extreme Gradient Boosting Outlier Detection (XGBOD). It presents a new concept called representation learning that studies systematic ways to discover representations for raw data without any human intervention. You will learn the outputs of unsupervised learning algorithms are fed to a supervised learning model to achieve model predictability.
Ch. 12 explains autoencoders and overall deep learning structure. You will apply autoencoders to detect anomalies. This chapter does not require readers to have deep learning background and teaches deep learning with a regression-friend approach.
The rarity of outliers implies the target will be extremely imbalanced in supervised learning models. Ch. 13 and 14 teach ten under- and over-sampling techniques to overcome the challenges.
Code examples
All code examples are in the GitHub folder. There is no need to type in the code.
273 pages, Paperback
Published January 3, 2023
1 person want to read
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
No one has reviewed this book yet.