List-columns in data.table

on YouTube ↗22m

" summary

Shows how to create and use list-columns in data.table to nest and unnest real examples (Star Wars characters with a per‑film revenue table), using the by argument, .SD and := (the walrus operator) plus purrr::map() to build per‑group list columns, then join and unnest back out. Bench::mark() comparisons to dplyr::group_nest() and tidyr::unnest() report lower memory use and faster runs as data and group count grows. A tiny R package tidyfast collects the helpers; use as.data.table vs setDT for copies.

AI-written summary · about

" chapters (5)

1. 0:00Intro
2. 7:07Example
3. 9:21Coding
4. 12:40Joining
5. 17:58Performance

" related talks

• 44mNeed for speed: the data manipulation versionSuzan Baert · satRdays Paris 2019
• 05mdplyr 1.1.0 Features You Can't Live WithoutDavis Vaughan · posit::conf(2023)
• 20mGive me the DataColin S Gillespie · posit::conf(2024)
• 18mduckplyr: Analyze large data with full dplyr compatibilityKirill Müller · posit::conf(2025)
• 51mDataFrame: Kotlin's Innovative Approach to Data StructuresRoman Belov · Devoxx Belgium 2024
• 18mdbtplyr: Bringing Column-Name Contracts from R to dbtEmily Riederer · posit::conf(2023)
• 05mWhy tidy data mattersGerard Walsh · satRday Johannesburg 2019
• 37mBenefits of columnar data layoutMarcin Stanisławski · JDD 2017