Week 2 Notes - Algorithm Desicison Making
Key Concepts Learned
What is An Algorithm
Definition: A set of rules of instructions for solving a problem or completing a task. Application: Criminal Justice/Housing&Finance/Healthcare.
Data analytics is Subjective
Every steps involves human choice embed human values and bias proxy don’t understand background and history
Group Challenge
Scenario: School enrollment assignment
Proxy: ZIP codes → stand-in for “assignment” Blind spot:ZIP codes are not a direct measure of individual circumstances. They can mask or undercount diversity, education levels within households, and financial status differences across families in the same area. Harm + Guardrail: Using ZIP code alone may reinforce existing segregation patterns, limit opportunities for disadvantaged students, and perpetuate inequities in school access.
Why Censs Data Matters
- Understanding community demographics
- Allocate government resources
- Tracking neighborhood changes
- Designing fair algorithm
Most policy analysis happens at - county level - census track level - block group level
Challenges & Concerns - Most computing can “re-identify” individuals from census data/privacy - solution: add noise to data set - privacy vs accuracy
Connections to Policy
Government using algorithm (Having limited budget and need to serve everyone) - Efficiency: process more cases faster - Consistency: same rules applied to everyone - Objectivity: removes human bias - Cost savings: fewer staff needed
Coding Techniques
- Install packages
- Clean data