LINEAR & NONLINEARSTARTER

Linear Regression

Best-fit line plus a causation audit. A big coefficient is not a big cause.

outlier pulls line
best-fit line ≠ cause · recreated schematic

How it works

  • Find the line minimizing squared misses.
  • Slope = expected change in Y per unit of X, holding the rest fixed.
  • R² says how much wiggle the line explains — not whether it is causal.
Use → Screen relationships, then hunt omitted variables and reverse causality.
Watch out → Omitted variables, reverse arrows, and outliers silently move the line.
Try it → Plot any two business metrics, add the line, then argue against it.
REDCAPE jobs → Reason · Predict · Communicate
Free sources → Coursera M3 (Categorical & Linear) • OpenIntro Statistics (CC BY-SA)
◍ Mentor soonSee REDCAPE playbook →

Attach to a lattice — combines with

Different families, shared jobs. Disagreement between them is where insight lives.

DISTRIBUTIONS
Normal Distribution
Bell-curve averaging: many small independent effects sum to a predictable center with thin tails.
SHARED JOBS: REASON · PREDICT · COMMUNICATE
DIFFUSION
Broadcast Model
One source reaches a fixed pool at a steady rate: fast early, then fading returns — the r-shaped curve.
SHARED JOBS: PREDICT · COMMUNICATE
LEARNING
Diversity Prediction Theorem
Crowd error equals average error minus diversity. Diverse models beat the lone genius.
SHARED JOBS: PREDICT · COMMUNICATE

More in Linear & Nonlinear

Categorical & Threshold Models
Buckets and cutoffs turn messy attributes into testable if-then rules.
Concavity & Diminishing Returns
Curved payoffs change optimal bets: diversify when concave, concentrate when convex.
← Entropy & UncertaintyCategorical & Threshold Models →