Getting Serious with Time Series

摘要： Time series—sequences of events or values that we can analyze to detect meaningful patterns—are a key element in data scientists’ toolkit.

Time series—sequences of events or values that we can analyze to detect meaningful patterns—are a key element in data scientists’ toolkit. From environmental studies to marketing, they play a crucial role across numerous stats-rich fields, and empower practitioners to draw insights about the past and make informed predictions about the future.

This week, we’ve selected four enlightening contributions that cover time-series forecasting from different angles. Regardless of your experience level, you’re likely to learn something new.

It’s important to get the basics right. If you’re taking your first steps in time-series analysis, Sameeha Afrulbasha ’s introduction to the topic is a great place to start. It offers clear definitions of fundamental concepts, focuses on ARIMA (autoregressive integrated moving average) models, and explains how the different components work together.

Bringing time series to life in Python. If you’re ready to roll up your sleeves and work with some real data, Leonie Monigatti makes the transition from theory to practice smooth and painless. This helpful resource provides all the detail—and code snippets—you’ll need to start manipulating and visualizing time-series data, and pays extra attention to sometimes tricky-to-handle datetime formats.

▲圖片來源：towardsdatascience.com

How to create a full time-series workflow. Once you feel comfortable with the building blocks of time-series forecasting, you may want to implement them into an end-to-end project. Marco Peixeiro ’s comprehensive guide will set you on the right path: it explains how to frame a time-series problem as a supervised learning problem, which then allows you to use any scikit-learn model

When deep learning meets time series. How can we apply the power of neural networks to forecasting tasks? Gabriele Orlandi explores the potential applications of cutting-edge models in the realm of time-series analysis, and shows how the latest research is shaking up classical approaches to prediction.

轉貼自： towardsdatascience.com

若喜歡本文，請關注我們的臉書 Please Like our Facebook Page： Big Data In Finance