Time series data is fun, but challenging. When ordering matters, your datasets get large, and timestamp precision differences can foil your merges, building reliable data processing pipelines requires a principled approach with the right tools. Enter data.table! Its expressive syntax will make your code powerful without sacrificing readability and its support for in-place operations will make your code super fast. Learn how to master time series data in data.table with this course!
Review of data.table
-This chapter provides an overview of all the cool things that make data.table perfect for working with time series data, including its multiple column-selection options, how to modify data.tables by reference, and calling functions by taking advantage of non-standard evaluation.
Getting time series data into data.table
-Ok, so you have some time series data and you believe me that data.table is great for it. Before you can test that, you need to convert your data! In this chapter, you'll learn how to convert from popular time series data formats into data.table.
Generating lags, differences, and windowed aggregations
-Like most other data, time series data you find in the wild are rarely suitable to directly start using in model training. In this chapter, you'll learn how to write powerful, expressive data.table code to implement a few common forms of time series feature engineering.
Case study: financial data
-It's time to put it all together! In this chapter you'll consider a real-world dataset of spot metal prices from the London Metal Exchange (LME). By the end, you'll know how to write reusable functions to perform common time series feature engineering tasks and you'll have experience using those functions to build a statistical model.