Subject
12 entries
Data Analysis
Bookmarks
Dash: self-learning data agent with 6-layer context
Dash is a self-learning data agent from Agno that grounds answers in 6 layers of context, inspired by OpenAI's in-house implementation. Each query improves the agent's performance via adaptive memory.
The Bell Curve Scar — Handicapping Passengers on the Unsinkable Ship
A statistical analysis of Titanic passenger survival rates examining how aggregated bell curve thinking obscures survival disparities by class, gender, and ticket price. Uses Titanic as a historical dataset to illustrate how summary statistics can mislead.
Data Analysis: The Hard Parts
Mikio Braun on the unglamorous hard parts of data analysis — bugs that look like insights, evaluation that requires ground truth you don't have, and reproducibility failures. A practitioner's counterweight to the hype around data science tools.
Intro to pandas Data Structures
Greg Reda's introduction to pandas data structures — Series, DataFrame, and Index — written in 2013 when pandas was still new enough to need a clear on-ramp. A canonical early tutorial that helped many data scientists learn the library.
An Introduction to Data Analysis — Statistics Done Wrong
Alex Reinhart's 'Statistics Done Wrong' introduction to data analysis — opens with the key insight that a p-value measures surprise, not correctness. A corrective for scientists trained in classical statistics who misinterpret their own results.
crush-tools: Custom Reporting Utilities for Shell
crush-tools is a Google-hosted collection of Unix command-line utilities extending the standard toolkit for custom reporting — field manipulation, aggregation, and transformation on tab-delimited files. Purpose-built for the kind of ad hoc data work that awk handles awkwardly.
Ad Hoc Data Analysis From the Unix Command Line
Wikibooks guide to ad hoc data analysis using Unix command-line tools — awk, sed, sort, uniq, cut, and friends. A practical reference for doing quick data exploration without loading a language or framework.
Crunching NYC Subway Data: A New Yorker's Busiest Stations
A data analysis post combining Node.js and SQL to parse MTA turnstile data and identify New York City's busiest subway stations. An early example of civic data journalism: public transit data as a lens on urban density and movement.
SADDLE — Scala Data Library
SADDLE is a Scala data manipulation library modeled after pandas and R's data.frame — bringing labeled, indexed data structures to the JVM. Aimed at quantitative analysts who wanted Python/R-style data manipulation within Scala's type system.
Data Analysis, Statistics, and Probability Overview
Annenberg Learner's Data Analysis, Statistics, and Probability course — a free online course covering statistical reasoning, data representation, and probability for educators and learners. A foundational resource for building statistical intuition before the MOOC era.
Data for Dummies: 6 Data Analysis Tools Anyone Can Use
GigaOM's roundup of data analysis tools accessible to non-engineers — covering self-service BI and visualization tools that put analysis in the hands of business users. A 2013 snapshot of the early self-service analytics movement.
I Analyzed the Chords to 1,300 Songs for Patterns
Hooktheory's analysis of chord progression patterns across 1,300 pop songs — using crowdsourced music theory data to find statistical regularities in harmony. A rare example of quantitative musicology producing genuinely surprising findings about song structure.