Skip to content
Faster Than Normal
Mathematics & Probability

Regression Analysis

Model #0787Category: Mathematics & ProbabilityDepth to apply:

By Updated 3 sources

5 min read

Contents

  1. 1. Core Idea
  2. 2. How to See It
  3. 3. How to Use It
  4. 4. Founders & Leaders
  5. 5. Connected Models
  6. 6. One Key Quote
  7. 7. Summary & Further Reading
Mathematics & Probability
Section 1

Core Idea

Regression analysis models the relationship between variables — identifying how changes in one or more inputs predict changes in an output. Linear regression fits a straight line; more complex forms capture curves, interactions, and multiple variables simultaneously. The power is prediction and decomposition: which factors actually drive the outcome, and by how much? The danger is confusing correlation with causation — a regression can show that two variables move together without proving one causes the other. In business, regression appears in pricing models, demand forecasting, customer lifetime value predictions, and any analysis that asks "what drives this metric?" The model teaches that quantifying relationships is essential, but trusting the model without understanding its assumptions is a recipe for confident error.

Get Faster Than Normal by email

Ideas from founders and companies.

Free newsletter. Unsubscribe anytime.

Or open the full subscribe page.

Section 2

How to See It

Growth
You're seeing it when your marketing team builds a model showing that email frequency correlates with revenue — but the regression ignores that the highest-frequency segment was already your most engaged customers. The relationship is confounded.
Product
You're seeing it when a feature-usage analysis shows that users who complete onboarding retain at higher rates. The regression identifies the association, but the question is whether onboarding causes retention or self-selects for already-motivated users.
Operations
You're seeing it when a demand-forecasting model uses last year's data to predict next year's sales, assuming the same variables and coefficients still hold. The regression is only as good as the stability of the underlying relationships.
Section 3

How to Use It

Build regression models to identify which variables drive your key metrics — but always question whether the relationship is causal or merely correlational. Check for confounding variables that might explain the apparent relationship. Validate models on out-of-sample data, not just the data used to build them. Use regression as a hypothesis generator, not a proof engine: it tells you what to investigate, not what to conclude.
Decision filter
"Does this regression show that X causes Y, or just that they move together — and what confounders might explain both?"
As a founder
Use regression to decompose your key metrics into their drivers — what actually moves retention, revenue, or conversion. But never ship a strategy based on a regression alone without asking what's being held constant, what's confounded, and whether the relationship would hold if conditions change.
Section 5

Founders & Leaders

Michael BloombergFounder, Bloomberg LP
Bloomberg built his fortune on the insight that financial professionals needed better tools to analyze relationships in market data — and regression analysis was at the core. The Bloomberg Terminal gave traders and analysts the ability to run regressions on asset returns, decompose risk factors, and model relationships between economic variables in real time. But Bloomberg himself understood the limit: models that fit past data perfectly often failed in new regimes. His business succeeded because it gave users the tools while emphasizing that judgment — not the regression output — makes the final call. Founders building data products or data-informed strategies should internalize the same principle: regression quantifies the relationship but doesn't guarantee it persists.
Section 7

Connected Models

Tension
Correlation vs Causation
Regression quantifies correlation but doesn't establish causation. The most common misuse of regression is treating its output as causal when it's merely associative.
Reinforces
Confounding Factor
Confounders are the hidden variables that make regression outputs misleading. Any regression without confounding checks can produce relationships that vanish when the true driver is controlled for.
Reinforces
Sensitivity Analysis
Sensitivity analysis tests how regression results change when assumptions shift. It's the stress test for regression models — revealing which findings are robust and which are fragile.
Section 8

One Key Quote

"All models are wrong, but some are useful."
— George Box
Section 11

Summary & Further Reading

Regression analysis quantifies the relationship between inputs and outputs — essential for understanding what drives your metrics. But it shows association, not causation. Always check for confounders, validate on out-of-sample data, and treat regression as a tool for hypothesis generation, not proof.
01
An Introduction to Statistical Learning — James, Witten, Hastie & Tibshirani (2013)
Book
Accessible treatment of regression and statistical learning methods with practical applications.
02
Mostly Harmless Econometrics — Angrist & Pischke (2009)
Book
How to use regression carefully to identify causal relationships, not just correlations.
03
Bloomberg by Bloomberg — Michael Bloomberg (1997)
Book
How Bloomberg built a data empire on the tools of quantitative financial analysis.

Why this matters next

mental modelsRegression to the Mean

Regression Analysis applied the Regression to the Mean mental model

mental modelsConfounding Factor

Regression Analysis applied the Confounding Factor mental model

mental modelsStatistical Learning

Regression Analysis applied the Statistical Learning mental model

mental modelsSignal vs Noise

Regression Analysis applied the Signal vs Noise mental model

mental modelsAll Models Are Wrong

Regression Analysis applied the All Models Are Wrong mental model

mental modelsCorrelation vs Causation

Regression Analysis applied the Correlation vs Causation mental model

Frequently asked questions

What is Regression Analysis?

Regression Analysis is a mental model used for better thinking and decision-making.

How do you apply Regression Analysis?

To apply Regression Analysis, identify situations where this framework is relevant, then use it as a lens to evaluate your options and decisions. The model is most useful when combined with other complementary mental models.

What category does Regression Analysis fall under?

Regression Analysis falls under the Mathematics & Probability category of mental models. Other models in this category can be found on the Mathematics & Probability hub page.

Why is Regression Analysis important?

Regression Analysis is important because it provides a structured way to think about problems that would otherwise be approached with intuition alone. Understanding this model helps you avoid common reasoning errors and make better decisions.

Continue exploring

BT

Mental model

Bayes Theorem

A mathematical framework for updating beliefs based on new evidence, proportiona

BT

Mental model

Black Swan Theory

Rare, unpredictable events with extreme impact that are retrospectively rational

CO

Mental model

Compounding

Small consistent gains accumulate exponentially over time — the most powerful fo

CC

Mental model

Correlation vs Causation

The critical distinction between two variables that move together and one actual

ER

Mental model

Ergodicity

The distinction between ensemble averages and time averages — what works across

EG

Mental model

Exponential Growth

Growth that accelerates proportionally to its current size, producing deceptivel

Get Faster Than Normal by email

Ideas from founders and companies.

Free newsletter. Unsubscribe anytime.

Or open the full subscribe page.

Popular Mental Models

First Principles ThinkingOccam's RazorCircle of CompetenceInversionConfirmation BiasSecond-Order ThinkingDunning-Kruger EffectSurvivorship BiasPareto PrincipleOpportunity Cost