A blog post warning about misuse of dplyr's ntile() function in R. The author, a tidyverse advocate, cautions that ntile() does not perform quantile-based binning as its name suggests and will produce statistically inaccurate results for data analysis, using a pharmacometric dataset example to illustrate the problem.
An analysis of the English indefinite articles 'a' and 'an' reveals that the correct choice depends on spoken sound rather than written letter. Of 32,455 words examined, only 129 required exceptions to the vowel-sound rule, such as 'unicorn' (consonant sound despite vowel letter) and 'hour' (vowel sound despite consonant letter).
A new DuckDB extension enables users to classify rows in CSV, Parquet files, or DuckDB tables using Jev. It processes approximately 1,000 rows in 10 seconds, offering faster performance than LLMs and better ergonomics than traditional classifiers for data analysis workflows.
A study analyzing 782 New York metro ZIPs over 26 years found that gentrification signals like coffee shop openings lag home price increases by about four years, making them poor predictors of future appreciation. A machine learning model ranking ZIPs by multiple factors achieved only modest one-year predictive power (+0.44) and performed worse than chance at ten-year horizons, suggesting real estate cycles are driven more by current market conditions than cultural markers.
A detailed benchmark compared Claude and ChatGPT's performance on data analysis tasks using live Zendesk support ticket data. Both models were tested identically across seven stages from discovery to self-audit, with results evaluated by a third Claude instance for factual accuracy against actual returned metrics.