WHAT THIS BOOK COVERS
This book organizes time series models and applications into six meticulously crafted parts. Each part equips you with the knowledge and skills to conduct time series forecasting and anomaly detection. The six parts are:
- Part 1: From Prophet to NeuralProphet
- Part 2: Getting Probabilistic Forecasts
- Part 3: Autoregressive-based Time Series Techniques
- Part 4: Tree-based Time Series Techniques
- Part 5: Deep into Deep learning-based Time Series Techniques
- Part 6: Transformer-based Time Series Techniques
WHY READ THIS BOOK?
In this book, you will find a wide coverage of methodologies, algorithms, and applications. Whether you’re a seasoned data scientist seeking to refine your expertise or a novice eager to embark on your analytical odyssey, this book offers a roadmap tailored to cater to diverse skill levels and objectives. You will emerge equipped with the proficiency and confidence to unravel intricate temporal patterns, harness predictive power, and unlock new horizons of insight across a myriad of domains, from finance and economics to healthcare and beyond. The writing style of this book is another selling point. Instead of taking a technique as given, this book first describes intuitions and then dives into detail. This book also gives a landscaping view from one idea to the consequent ideas. With the real-world data cases in this book, you will gain a deeper understanding of how time series techniques are applied in diverse domains.
TABLE OF CONTENTS
- Preface
- Introduction
- Prophet for business forecasting
- Tutorial I
- Tutorial II
- Change Point Detection in Time Series
- Monte Carlo Simulation for Probabilistic Forecasting
- Quantile Regression for Probabilistic Forecasting
- Conformal Predictions for Probabilistic Forecasting
- Conformalized Quantile Regression for Probabilistic Forecasting
- Automatic ARIMA!
- Time Series Data Formats Made Easy
- Linear Regression for Multi-period Probabilistic Forecasting
- Feature Engineering for Tree-based Time Series Models
- Two Primary Strategies for Multi-period Time Series Forecasting
- Tree-based XGB, LightGBM, and CatBoost Models for Multi-period Probabilistic Forecasting
- The Progression of Time Series Modeling Techniques
- Deep Learning-based DeepAR for Probabilistic Forecasting
- Application — Probabilistic Predictions for stock prices
- From RNN to Transformer-based Time Series Models
- Temporal Fusion Transformer for Interpretable Time Series Predictions
- Lag-Llama for Time Series Forecasting
WHAT YOU GET IN THE BOOK
- Learning time series techniques comprehensively in a short period of time
- A roadmap from the classical techniques to modern time series forecasting
- Applying forecasting for resource planning and anomaly detection
- Mastering time series Python libraries
- Model interpretability
- Model evaluation metrics
- Hands-on example code
- Cheat Sheets
Leveraging his 20+ years of teaching experience at Columbia University and Boston University and other universities, Kuo distills complex concepts into a comprehensive and accessible format.
The book is a testament to Kuo’s deep understanding of time series analysis and its applications in predictive analytics and anomaly detection. This book equips readers with the necessary skills to tackle real-world challenges. It is particularly valuable for those seeking a career change into data science.
Kuo provides a detailed exploration of both traditional and cutting-edge techniques. Kuo integrates discussions on neural networks and other advanced algorithms, reflecting the latest trends and developments in the field. This ensures that readers are not only learning established methods but are also prepared to engage with the most current and innovative techniques in data science.
The book’s clarity and accessibility are enhanced by Kuo’s engaging writing style. He successfully demystifies complex mathematical and statistical concepts, making them approachable without sacrificing rigor.