Full Citation
Title: Predicting Migration Outcomes During the Great Depression: Using Decision Trees and Full Count US Census Data for 1930-40
Citation Type: Dissertation/Thesis
Publication Year: 2023
ISBN:
ISSN:
DOI:
NSFID:
PMCID:
PMID:
Abstract: The Great Depression is one of the worst social, financial, economic, and environmental disasters in modern American history. Economic and climate distress provoked a large internal migration of people. This thesis attempted to glean predictions about the migratory decisions in the 1930s using decision trees utilizing linked full count US census data from 1930 and 1940. Five different models were employed to assess the 52 million linked records. Using variables from one census year alone provided poor classification rates for the migrants but near perfect rates for the non-migrants. Using a combination of all variables across both years provided the highest success in classifying both migrants and non-migrants. However, this was achieved by successive splitting on the basis of a location identifier that combines county and state across the two years. Excluding the location identifier for 1940 alone did produce better success than single years but still misclassified a large proportion of the migrants. The analysis conclusively pointed to the excessive reliance on geographic variables for the decision tree to make splitting decisions, contrary to the expected demographic and occupation details. The reliance potentially hints at spatial relationships or local shocks, that are not captured in the census data. In conclusion, it appears that the census data is missing some attribute(s) that makes people arrive at the difficult decision of migrating.
Url: http://rave.ohiolink.edu/etdc/view?acc_num=case1673036935940106
User Submitted?: No
Authors: Kesavan, Pooja
Institution: Case Western Reserve University
Department: Mathematics
Advisor:
Degree:
Publisher Location:
Pages: 1-61
Data Collections: IPUMS USA, IPUMS USA - Ancestry Full Count Data
Topics: Migration and Immigration
Countries: