Cody Arlie ReedResearch + data

Research portfolio

Projects

Research on racial segregation, occupational knowledge, and intergenerational mobility.

Cities and mobility

Racial segregation at home and across daily travel.

Paper · 2026

Segregation across the day

How does racial segregation change across hours and days? I developed a method that combines mobile-phone locations, Census data, and high dimensional matrix multiplication to assess three theoretical perspectives on racial segregation dynamics.

  • Segregation estimation
  • Matrix computation
  • R and C++
Read the project

From location records to racial segregation

  1. Mobile-phone locations + Census dataAttach the racial composition of each phone’s inferred home tract.
  2. Weight nearby observationsMy matrix formulation combines demographic data with distances in space and time.
  3. Estimate segregation each hourCompare local racial composition with the commuting zone’s overall composition.
A continuous measure of racial segregation across space and time.

Method and interactive dashboard

Locating sharp racial boundaries

Where do sharp racial boundaries occur within a city? Taking inspiration from image edge detection, I developed a measure that accounts for the strength and direction of racial gradients and the local presence of both groups.

  • Metric construction
  • Spatial gradients
  • Visualization
Read the project

How the measure works · Synthetic population fields

Boundary scoreSynthetic spatial surface: Boundary score. Darker burgundy indicates higher values.0–1 relative to this example’s maximum
Two overlapping population fields and a constant background population. The boundary score is highest where both groups are present and their shares change sharply in opposite directions.

Work and occupational knowledge

Occupational knowledge, class reproduction, and text classification.

Population Association of America · 2024

Knowledge of parents’ work

Does knowing about a parent’s work help children enter the same occupational class? I developed a content-weighted semantic similarity measure to compare students’ and parents’ descriptions of the same parent’s job.

  • Word embeddings
  • Metric construction
  • Survey-weighted regression
Read the project

Synthetic example · a parent’s teaching job

Parent-survey description
“teaches mathematics; plans lessons; grades assignments.”
Student’s description of the same job
“explains arithmetic; prepares lessons; checks assignments.”
Both descriptions refer to teaching, although the wording differs. Word embeddings let me compare these accounts of the same job.

Research project

Classifying jobs from short descriptions

Can a short description of a job identify its occupation? I trained an ensemble of five TensorFlow neural networks using Word2Vec representations to rank 120 occupation groups.

  • Text classification
  • Word2Vec
  • TensorFlow
Read the project
  1. 01 · InputJob title and duties

    Short descriptions in everyday language

  2. 02 · RepresentationWord2Vec

    A 256-value representation of each description

  3. 03 · ClassificationFive neural networks

    TensorFlow models rank 120 occupation groups

  4. 04 · OutputTwo suggested codes

    Ranked predictions and model scores for review

From survey text to ranked occupation codes