An API for Everything There Is to Know About Packages

Understanding Tail Analysis in Financial Markets

A practical guide to tail analysis in financial markets using R, comparing Gaussian (thin-tailed) and Student-t (fat-tailed) distributions to classify price movements as either information-driven or liquidity-driven. The workflow fetches gold futures data (GC=F) via tidyquant, fits both distributions using MASS::fitdistr(), compares log-likelihoods to determine the dominant regime, and presents results in a color-coded audit table built with the gt package. A volume impact metric is also computed to highlight whether abnormal trading activity amplifies price changes.

Rethinking Validation for Spatial Machine Learning: Takeaways from the Talk

A keynote talk and workshop at ML4EO 2026 challenged three common assumptions in spatial machine learning validation: that models can predict everywhere, that one validation approach fits all, and that all validation points are equally weighted. The talk introduced 'prediction-domain adaptive evaluation', which involves defining the prediction domain first, constructing validation folds that reflect it (using tools like Area of Applicability, Local Point Density, and kNNDM cross-validation), and summarizing performance weighted by prediction condition prevalence. The workshop provided practical R workflows implementing these ideas. The core message: spatial ML evaluation must answer not just 'how accurate is the model?' but 'where can the model be trusted?'

A New Guide: Organizing Events for First-time Contributors

rOpenSci has published a free guidebook titled 'From User to Contributor: Organizing Events for First-Time Contributors', funded by a NumFOCUS Small Development Grant. The guide documents two event formats designed to lower barriers for first-time open source contributors: mini-translathons (short coworking sessions focused on translation and localization) and mini-hackathons (short online sessions for fixing bugs, improving code, or updating documentation). The guide covers timelines, community calls, event formats, pilot event results, and includes communication templates. It is intended to be reusable by communities beyond rOpenSci.

stick function for the EDA in time series

A new stick function (Seasonal, Trend, Irregular Contribution Kit) has been added to the greybox package for both R and Python. Based on Hans Levenbach's STI classification, it uses ANOVA to quantify the strength of seasonal, trend, and irregular components in a time series. Applied to the classic AirPassengers dataset, it reveals that trend accounts for ~86% of variability. The function supports multiple seasonal components and may be extended to include external regressors in the future.

Set Working Directory in R: setwd() & RStudio GUI Guide

A guide on how to set the working directory in R using the setwd() function or the RStudio Session menu GUI. Covers getwd() for checking the current directory, handling Windows path errors, and using the here() package for more robust project-relative paths, particularly useful for dissertation projects.

Unlearning the Tidyverse

Một nhà phát triển R lâu năm quyết định từ bỏ Tidyverse sau 8 năm sử dụng do trải nghiệm tiêu cực khi đóng góp vào package 'httpuv'. Anh ấy dự định chia sẻ cách thực hiện các tác vụ xử lý dữ liệu phổ biến bằng base R thay thế.

Lập trình viên R nên đọc bài này để hiểu cách chuyển đổi từ các công cụ mạnh mẽ của Tidyverse sang cơ sở hạ tầng cơ bản của R, giúp tiết kiệm thời gian và tránh những trải nghiệm thất vọng khi đóng góp vào cộng đồng mở.

EFA vs CFA: Key Differences Between Exploratory & Confirmatory Factor Analysis (R)

A comprehensive side-by-side comparison of Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA) with full R code examples. EFA uses the psych package to discover unknown factor structures from data, while CFA uses lavaan to test pre-specified models. Covers when to use each method, assumptions, sample size requirements, model fit indices (CFI, TLI, RMSEA, SRMR) with accepted thresholds, common mistakes to avoid, and how to correctly sequence both methods in dissertation research using independent samples.