Week-2-notes

Algorithmic Decision Making & Census Data

Inputs: Synonyms: Independent variables, “x”’s, predictors, features

Outputs: Synonyms: dependent variable

Government

-            Idea is that using data will result in more consistent and unbiased output

-            Reasons for use: Efficiency, consistency , objectivity, cost savings

1)       Data Science

2)       Data Analytics (MUSA)

3)       Machine Learning 

4)       AI

Subjectiveness

-            Data cleaning decisions

-            Data coding or classification

-            Collection

-            How you interpret results

-            What variables you use

Census Data Foundations

·       9 basic questions (age, race, sex, etc.)

·       Everyone counted every 10 years

·       Constitutional requirement

·       Determines political representation

American Community Survey (ACS)

·       3% of households surveyed annually

·       detailed questions (income, education, housing costs, etc)

o   replacement of the same “long form” in 2005

·       Areas: of only 65,000 people.

·       Pretty small sample but you get them every year. Only aggregate level

·       5-year estimates

o   Take all 1-year and combine together including census tracts (1500-1800 people). Most accurate we can get at the smallest geography and we consider them a neighborhood.

§  It is changed for redistricting

§  Frustrating for us when redistricting happens

o   Most reliable data, largest sample

Tidycensus

-            This package helps organize the census data