: Traditional models like ARIMA and Exponential Smoothing are presented as robust baselines, especially for smaller datasets where complex models might overfit.
Nielsen argues that time series analysis is often underrepresented in standard data science toolkits despite its ubiquity. The book emphasizes that temporal data is fundamentally different from cross-sectional data because of:
: Nielsen spends significant time on "data munging"—cleaning, handling missing values, and addressing outliers. She notes that "fancy techniques can't fix messy data". Practical Time Series Analysis - Aileen Nielsen...
The book is structured to lead readers through the full lifecycle of a time series project:
Aileen Nielsen’s Practical Time Series Analysis stands out as a multidisciplinary guide that fills a significant void in modern data science literature. While many textbooks focus strictly on classical econometrics or purely on deep learning, Nielsen offers a comprehensive pipeline that integrates both worlds for real-world applications like healthcare, finance, and the Internet of Things (IoT). : Traditional models like ARIMA and Exponential Smoothing
: The guide introduces non-linear approaches such as Random Forests , XGBoost , and Deep Learning (LSTMs, CNNs, and Transformers) for capturing complex temporal patterns.
: Unlike general regression, the time variable does not repeat, making forecasting an extrapolation challenge. She notes that "fancy techniques can't fix messy data"
For those looking to dive in, the book provides a "multilingual" experience, alternating between and R code examples.