home.social

#data-analysis — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #data-analysis, aggregated by home.social.

fetched live
  1. Digital Preservation at the University of Cambridge Libraries and Archives: Using Data Analysis to Untangle a 75TB Digital Preservation Problem. “For over two decades starting in 2000, the Streaming Media Service (SMS) served as the primary platform for managing and delivering audio and video content across the University. When the service reached its end-of-life in late 2025, the entire system […]

    https://rbfirehose.com/2026/08/21/digital-preservation-at-the-university-of-cambridge-libraries-and-archives-using-data-analysis-to-untangle-a-75tb-digital-preservation-problem/
  2. A new method ("Bonsai") for analysis of high-dimensional data uses Bayesian statistics to construct a two-dimensional tree representation with data points as leaves and branch lengths corresponding to the probability of transformation paths from one to another. The code is in Python and is licensed CC-BY-NC.

    Summary: phys.org/news/2026-08-unseen-t

    Original paper: nature.com/articles/s41587-026

    #Science #DataAnalysis #Multivariate #Python

  3. Three ways to get answers out of business data without writing Python:

    Send the dataset straight to a model and ask what the patterns are.
    Connect the database and query it in plain language instead of SQL.
    Train a classifier or regressor on a CSV, then run predictions at no per-request cost.

    Plenty of teams reach for the third when they needed the first.

    aiappsapi.com/ai-analytics/

    #AI #DataAnalysis

  4. Variety: ABC Unveils New Twist for Live-Streamed News: Shows That Respond to Viewer Queries, Needs. “During a recent live-stream of ‘Searched’, a relatively new entry to the schedule of ABC News Live, Benitez, clad in bright white sneakers and a short sleeve shirt, offered followers different angles on top news stories — the ones they were asking about on their own. ‘Searched’ uses data from […]

    https://rbfirehose.com/2026/08/18/abc-unveils-new-twist-for-live-streamed-news-shows-that-respond-to-viewer-queries-needs-variety/
  5. Computer Weekly: Natural History Museum’s ‘living laboratory’ logs 11 million environmental records in first year. “The Natural History Museum’s (NHM’s) ‘living laboratory’ garden has collected more than 11 million environmental data records in its first year, feeding readings from more than 50 sensors into the Museum’s Amazon Web Services (AWS)-powered Data Ecosystem.”

    https://rbfirehose.com/2026/08/18/computer-weekly-natural-history-museums-living-laboratory-logs-11-million-environmental-records-in-first-year/
  6. Alright, future engineers!
    **Degrees of Freedom (DF):** The number of values in a calculation that are free to vary.
    Ex: For 'n' data points & 1 mean, DF = n-1.
    Pro-Tip: Vital for choosing the right stat test & interpreting results (e.g., t-test)!
    #Stats #DataAnalysis #STEM #StudyNotes

  7. Discovering my own music tastes
    A side-project and data-driven journey to confirm what I already knew, but to also show me how things change overtime.

    ericof.com/en/posts/2026/music

  8. Alright, future engineers!
    **Standard Deviation:** Measures how spread out data points are from the average (mean).
    Ex: Small Std Dev means data are tightly clustered around the mean.
    Pro-Tip: It quantifies data *variability* – essential for understanding consistency or risk in any dataset!
    #Statistics #DataAnalysis #STEM #StudyNotes

  9. Ah, the riveting saga of "model card benchmarks" 🤔—because nothing screams excitement like sifting through 30 flavorless data points! 📊 The article tries to spice up the nerd parade with rankings and trends, but let's face it, it's basically the PowerPoint presentation you slept through in college. 🎉🚀
    koutian.is-a.dev/benchmark-rad #modelcardbenchmarks #dataanalysis #techhumor #boringpresentations #nerdculture #HackerNews #ngated

  10. RT @glenngabe: Möchtest du AI-Mode-Prompts für deine Website sehen? - In meinem neuesten Beitrag erkläre ich Schritt für Schritt, wie du alle Query-/Landing-Page-Daten über die GSC-API exportierst und Claude nutzt, um zu bewerten und zu organisieren, was deiner Meinung nach AI-Mode-Query-Daten sind. Sei gewarnt, in den Daten stecken einige überraschende Prompts. Am Ende verfügst du über eine Claude-Skill, die du jederzeit wieder verwenden kannst, um die Analyse erneut durchzuführen. Viel Spaß. :) Glenn Gabe (@glenngabe) Seit Barry's Beitrag letzte Woche über AI-Mode-Queries, die in GSC auftauchen, habe ich über einen saubereren Weg nachgedacht, um mehr dieser Daten (sowie zitierte URLs) zu erhalten. Du kannst meinen Beitrag lesen, um dies heute umzusetzen. Sei gewarnt, da stecken einige überraschende Prompts drin :) - Wie man AI-Mode-Queries über die Search Analytics API, Analytics Edge und Claude Cowork findet und organisiert Indem du alle Query- und Landing-Page-Kombinationen über die Search Analytics API exportierst und dann Claude Cowork nutzt, um AI-Mode-Queries sichtbar zu machen, kannst du am Ende ein sauberes Spreadsheet mit möglichen Prompts sowie Claudes Begründung erhalten. Du erhältst auch die Prompts und die Landing Pages, die in AI Mode zitiert wurden. Es ist nicht perfekt, aber es ist interessant, den Prozess durchzuarbeiten. gsqi.com/marketing-blog/how-… — nitter.net/glenngabe/status/20

    mehr auf Arint.info

    #AI #Claude #DataAnalysis #Marketing #SearchAnalytics #SEO #arint_info

    https://x.com/glenngabe/status/2087957153793659296#m

  11. Alright, future engineers!

    **Standard Deviation (SD):** Measures how spread out data points are from the mean. Ex: Test scores avg 75. SD=5 means most scores are near 75. SD=15 means scores vary widely. Pro-Tip: A small SD indicates data consistency/precision!
    #Stats #DataAnalysis #STEM #StudyNotes

  12. ICYMI: Microsoft Clarity card ranks which AI operators scrape most and refer least: Sample card displays a 6,000:1 ratio against 41 referrals, then links straight into filtered session recordings. Free to use, but only with a CDN connected. ppc.land/microsoft-clarity-car #MicrosoftClarity #AIoperators #DataAnalysis #WebAnalytics #DigitalMarketing

  13. FYI: Stanford study finds 5% of IP addresses send 55% of all web requests: Fewer than 0.2% of domains hold a unique IPv4 address, and 44% of client IPs carry two or more user agents. Frequency caps and fraud filters read that signal. ppc.land/stanford-study-finds- #StanfordStudy #IPAddresses #WebTraffic #DigitalMarketing #DataAnalysis