Part 4 case study

Lottery ZIP Psychographics: How Neighborhoods Play

A NYC ZIP-level segmentation built from NY Lottery baseline behavior. The model ignores demographics while it learns the behavior space, then uses income, education, race, ethnicity, population, and retailer availability to interpret the segments afterward.

Behavior Space

PCA scores from 35 lottery behavior features

Each dot is an active NYC ZIP. Color is the final k-means segment.

Step 1

Do not cluster on dollars.

Raw lottery sales mostly follow population and retailer availability. The case starts by separating scale from play style.

Step 2

Compress behavior.

PCA turns correlated product, channel, add-on, timing, entropy, and habit measures into readable coordinates.

Step 3

Rotate for meaning.

Factor analysis names latent play modes: daily habit, rapid convenience, jackpot add-ons, and incidental variation.

Step 4

Profile the clusters.

K-means on behavior scores gives three segments. Demographics explain them only after the clusters are formed.

PCA Behavior Map

x = daily habit/social routine, y = jackpot/add-on intensity

Behavior first, labels second.

The axes and clusters use product mix, channel context, add-on rates, timing, concentration, entropy, volatility, and habit signals. No demographics enter the fit.

Segments are lenses, not causes.

A ZIP placed in the daily-routine segment is behaviorally similar to other daily-routine ZIPs. That does not say why residents play that way.

Segment Profile

cluster means, NYC active ZIPs

Borough Strip

grouped ZIPs, not polygon geometry

Vertical position follows the first behavior component. Point size follows lottery sales.

PCA and Loadings

standardized behavior features only

Why Raw Sales Are Not a Segment

correlations in the analysis sample

Factor Analysis Readout

rotated loadings

PCA gives the compact map. Factor analysis gives a language for the latent play styles behind the observed variables.

Demographics After Scoring

not used in the fit

Method

gallery version

Caveats

read before teaching