Blog

R, Stats and Data Science related blog posts

Fast spatial data matching in R

Exploring different solutions to match locations based on their geographical distance
R
DuckDB
Spatial
GIS
Big Data
SQL

This post showcases various solutions to efficiently match unknown locations to known ones by their geographical proximity (using lat/long coordinates), on a dataset with millions of entries.

Jun 18, 2022
42 min

MCMC for ‘Big Data’ with Stan

Faster sampling with CmdStan using within-chain parallelization
R
Statistics
ML
Bayesian Modeling
Stan
Big Data

This post is an extension (and a translation to R) of PyMC-Labs’ benchmarking of MCMC for “Big Data”.

The Stan code was updated to use within-chain parallelization and compiler optimization for faster CPU sampling. Stan was able to achieve similar sampling speeds as PyMC’s JAX + GPU solution, purely on CPU.

Jun 5, 2022
10 min

Data wrangling with data.table and the Tidyverse

Common data wrangling operations with both data.table and the Tidyverse.
R
Tidyverse
data.table

This post showcases various ways to accomplish most data wrangling operations, from basic filtering/mutating to pivots and non-equi joins, with both data.table and the Tidyverse (dplyr, tidyr, purrr, stringr).

May 19, 2022
45 min

Bayesian Rock Climbing Rankings

With R and Stan
R
Statistics
ML
Bayesian Modeling
Stan

This post is a transposition to R of Ethan Rosenthal’s blog post on modeling Rock Climbing route difficulty using a Bayesian IRT (Item Response Theory) model.

The original Stan code was updated to use within-chain parallelization and compiler optimization for faster CPU sampling.

Several data processing solutions are showcased, using either data.table or dbplyr (with a DuckDB backend), with timings to compare their speed.

Apr 19, 2022
16 min
No matching items
Back to top