展示HN：掌握现代时间序列预测——Python实用指南
我正在编写一本关于Python时间序列预测的实用指南，名为《掌握现代时间序列预测》。其目标是弥合理论与实践之间的差距。书中涵盖了经典统计模型（如ARIMA、SARIMA、Prophet）以及现代机器学习/深度学习方法（如N-BEATS、Transformers、Temporal Fusion Transformer）。
代码示例使用了Python库，如statsmodels、scikit-learn、PyTorch和Darts，书中重点关注实际工作流程：处理杂乱数据、特征工程、模型选择和评估。我写这本书是因为在寻找既实用又最新的预测资源时遇到了困难，尤其是对于应用机器学习的从业者来说。
您可以在这里找到这本书：<a href="https://valeman.gumroad.com/l/MasteringModernTimeSeriesForecasting" rel="nofollow">https://valeman.gumroad.com/l/MasteringModernTimeSeriesForec...</a> <a href="https://leanpub.com/mastering_modern_time_series_forecasting" rel="nofollow">https://leanpub.com/mastering_modern_time_series_forecasting</a>
欢迎任何从事时间序列预测或使用Python机器学习工具的人提问或反馈。
查看原文
I’ve been working on a practical guide for time series forecasting in Python, called Mastering Modern Time Series Forecasting.
The goal is to bridge the gap between theory and implementation. It covers both classical statistical models (ARIMA, SARIMA, Prophet) and modern machine/deep learning approaches (N-BEATS, Transformers, Temporal Fusion Transformer).
The code examples use Python libraries like statsmodels, scikit-learn, PyTorch, and Darts, and the book focuses on real-world workflows: messy data, feature engineering, model selection, and evaluation.
I wrote this after struggling to find forecasting resources that were both practical and up-to-date — especially for applied ML practitioners.
You can find the book here: <a href="https://valeman.gumroad.com/l/MasteringModernTimeSeriesForecasting" rel="nofollow">https://valeman.gumroad.com/l/MasteringModernTimeSeriesForec...</a> <a href="https://leanpub.com/mastering_modern_time_series_forecasting" rel="nofollow">https://leanpub.com/mastering_modern_time_series_forecasting</a>
Happy to answer questions or hear feedback from anyone working with time series forecasting or Python ML tools.