agentsclimarketplace

Regression analysis

Skill Amey-Thakur/AI-SKILLS/skills/data-science/regression-analysis

Plug-and-play skills and prompts for every AI coding agent

Install
npx -y skills add Amey-Thakur/AI-SKILLS --skill regression-analysis

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • 19 days oldThe repository was created 19 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
  • 4 stars4 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Fit and interpret regression models for insight, reading coefficients, fit, and caveats honestly rather than as causal truth. Use when using regression to understand relationships in data, not to build a predictive model.

SKILL.md

3.4 KB, as published. Nobody here has run it

Regression analysis

Regression is the workhorse of analytics: it quantifies how outcomes relate to inputs and lets you say "holding other things equal, X is associated with this much Y". Its power is also its trap, because a coefficient looks like a causal effect and usually is not. Interpreting it honestly is the skill.

Method

  1. Pick the form for the outcome. Linear regression for continuous outcomes, logistic for binary (odds/probability), Poisson-family for counts. Using linear regression on a 0/1 outcome or ignoring the outcome type gives nonsense predictions and misleading coefficients.
  2. Read coefficients as "holding others constant". Each coefficient is the association between that predictor and the outcome with the other included variables held fixed. This "controlling for" is only as good as the variables you included; an omitted confounder biases the coefficients you have (see correlation-causation). State the direction, size, and units, not just significance.
  3. Judge fit and the residuals, not just R-squared. R-squared says how much variance is explained but nothing about correctness; a high R-squared can hide a mis-specified model, and a low one can still carry a real relationship. Plot the residuals: patterns reveal nonlinearity, heteroscedasticity, and outliers the summary numbers miss.
  4. Report uncertainty on every coefficient. Coefficients are estimates with confidence intervals; a "significant" coefficient with a wide interval is weakly known. Give the interval, and remember that with enough data, tiny, unimportant effects become significant (see statistical-inference).
  5. Check the assumptions and the specification. Multicollinearity (correlated predictors make individual coefficients unstable and uninterpretable), the right functional form (linear in the right variables, interactions where they matter), and influential outliers. A model that violates its assumptions produces confident, wrong coefficients.
  6. Resist the causal leap. A regression coefficient is an association, not the effect of intervening, unless the data came from an experiment or you have carefully handled confounding. "Controlling for X" in a regression is not the same as an experiment; do not tell stakeholders a coefficient is what will happen if they change the input (see correlation-causation, experiment-analysis).

Boundaries

  • Interpretive regression (understanding relationships) and predictive modeling (forecasting an outcome) optimize for different things; a model built for insight is judged on honest coefficients and caveats, one for prediction on held-out accuracy (see model-evaluation, ml-problem-framing).
  • Regression on observational data supports "associated with", not "causes"; the coefficient's causal interpretation requires a design (experiment, natural experiment) the regression alone does not provide.
  • Complex relationships (strong nonlinearity, interactions, high dimensionality) may exceed what a readable regression captures; there the tradeoff is interpretability versus a flexible model you cannot explain as simply.

Gives 0 of the 12 instructions most data analysis skills give

Counted across 286 of the 286 authors here whose files we hold, read 2026-08-06

  • use excel formulas instead of hardcoded calculated valuesin 35 of 286, across 7 files
  • match existing template conventions when modifying filesin 35 of 286, across 7 files
  • document sources for all hardcoded valuesin 35 of 286, across 7 files
  • write minimal concise python codein 35 of 286, across 7 files
  • place all assumptions in separate assumption cellsin 32 of 286, across 5 files
  • apply industry-standard color coding to financial modelsin 31 of 286, across 5 files
  • format years as text stringsin 30 of 286, across 3 files
  • recalculate formulas using recalc.py after modificationsin 30 of 286, across 3 files
  • format negative numbers using parenthesesin 30 of 286, across 3 files
  • fix all identified formula errors before finishingin 27 of 286, across 1 file
  • use colorblind-safe palettesin 19 of 286, across 12 files
  • Name tests after the prevented bugin 13 of 286, across 8 files

Said here and by no other author read

  • pick regression form based on outcome type
  • read coefficients as holding other variables constant
  • state coefficient direction size and units
  • plot residuals to check model fit
  • check for multicollinearity and influential outliers
  • state model assumptions and specification limits

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.