About the book
WHAT'S NEW IN THE 2ND EDITION
Comprehensive and in-depth coverage: The second edition offers a wide range of techniques for anomaly detection. It ensures a well-rounded understanding of the foundational concepts, making it suitable for beginners and seasoned professionals alike. Each chapter builds on the previous one, creating a cohesive and comprehensive learning experience.
Enhanced explanations and visual presentations: We’ve enriched this edition with a wealth of visualizations. These enhancements are designed to clarify complex concepts. You’ll learn not just the ‘how’ but also the ‘why’ behind each method.
Supervised and unsupervised learning techniques: We expand the second edition to cover the advanced machine learning techniques for anomaly detection. On the unsupervised learning part, you will be equipped with a wide range of modern techniques to discover new patterns. On the supervised learning part, you will gain a deeper understanding into grid-search, hyperparameter tuning, regularization, under-sampling and over-sampling techniques. You will use the code examples to build a suite of high-power machine learning models.
200+ data science Q&A: One of the standout features of the second edition is the inclusion of the Q&A sections in each chapter. By working through these questions, you’ll be well-prepared to showcase your expertise and secure your next job opportunity.
TABLE OF CONTENTS
Chapter 1: Introduction
Chapter 2: Histogram-Based Outlier Score (HBOS)
Chapter 3: Empirical Cumulative Distribution-based Outlier Detection (ECOD)
Chapter 4: Isolated Forest
Chapter 5: Principal Component Analysis (PCA)
Chapter 6: One-Class Support Vector Machine (OC-SVM)
Chapter 7: Gaussian Mixture Model (GMM)
Chapter 8: K-nearest Neighbors (KNNs)
Chapter 9: Local Outlier Factor (LOF)
Chapter 10: Cluster-Based Local Outlier Factor (CBLOF)
Chapter 11: Autoencoders
Chapter 12: Supervised Learning Primer
Chapter 13: Regularization
Chapter 14: Sampling Techniques for Extremely Imbalanced Data
Chapter 15: Representation Learning for Outlier Detection
Learn with Chris Kuo
Chris Kuo is a data scientist and an adjunct professor with 23+ years of experience. He led various data science solutions including customer analytics, health data science, fraud detection, and litigation analytics. He is also an inventor of a U.S. patent. He has worked at several Fortune 500 companies in the insurance and retail industries. In addition to teaching at Columbia University, he has taught courses in time series forecasting, mathematical finance, economics, and management at Boston University, University of New Hampshire, and Liberty University. Chris Kuo received his Ph.D. in Economics from SUNY at Stony Brook and his B.S. in Nuclear Engineering from National Tsing-Hua University, Taiwan. He and his wife live in New York, New York.