Time Series with PyTorch

Time series is far more than fit-predict forecasting. Real mastery comes from intuition and is built through experimentation. Walk the full range with two practitioners: forecasting, conformal prediction, transfer learning, and beyond. Key FeaturesGrasp core concepts through clear explanations that build genuine understanding rather than surface familiarityWork with realistic datasets and develop the judgement to choose the right approach for your problemProgress from neural network fundamentals to advanced techniques across a full range of time series challenges.Book Description“The first book since mine to cover this much ground, with a deep learning-first perspective few match. For a holistic treatment of time series, from forecasting to classification, clustering, and emerging architectures, this is the book for you.”— Manu Joseph, author of Modern Time Series Forecasting with Python, creator of PyTorch TabularNeural networks are powerful for time-series forecasting, but using them well takes practical experience and a clear grasp of architectures, training strategies, and evaluation methods. This book brings them together practically.Starting with PyTorch fundamentals, you’ll build networks from scratch, then progress through recurrent networks, attention, and transformers to N-BEATS, N-HiTS, and the Temporal Fusion Transformer. You’ll also learn robust hyperparameter tuning, conformal prediction for uncertainty estimation, and reliable evaluation.Unlike most forecasting books, it covers overlooked topics: transfer learning across series collections, synthetic data generation with diffusion models, and self-supervised representation learning. Later chapters cover classification, clustering, anomaly detection, and embeddings for large-scale modeling.Theory is reinforced through experimentation and implementation, so you can apply these methods confidently to real-world problems.What you will learnBuild, train, and evaluate neural networks for time series using PyTorch and PyTorch Lightning. Tune models with Bayesian optimisation and validate them with suitable metrics and strategies.Progress from feedforward and recurrent networks to transformers and models such as N-BEATS, N-HiTS, and TFT.Learn how global models use cross- and transfer learning across many series.Generate synthetic series and representations with diffusion and self-supervised methods.Apply modern approaches to classification, clustering, and anomaly detection.Who this book is forThis book is for data analysts, scientists, and students who want to know how to apply deep learning methods to time-series forecasting problems with PyTorch for real-world business problems.While the book assumes some understanding of statistics and modeling, you won’t need in-depth knowledge of time series to follow along. Some familiarity with Python is important, but we do not assume any prior knowledge of PyTorch.The main goal of this book is to be accessible to those with little or no experience with deep learning methods in time series.