Multigenerational Income Mobility Data for Predictive Modeling

Aggregating 1,365 observations for a predictive modeling exercise on income mobility across three generations using Panel Study of Income Dynamics (PSID) data. The task is to predict log income in generation 3 using log incomes from prior generations, education levels, race, and sex. It includes separate learning and holdout files for model training and evaluation.

Use Cases

• Predict g3_log_income using g1_log_income and g2_log_income as primary features in a regression model.
• Analyze the relationship between education levels (coded as less than high school, high school, some college, or 4+ years of college) in each generation and income mobility outcomes.
• Assess the predictive influence of race (as reported by the grandparent) and sex of the respondent in generation 3 on log income.
• Build a model using the 1,365 observations in learning.csv to generate predictions for the corresponding holdout_public.csv file.

Strengths

• Contains 1,365 observations for model training with a corresponding holdout set of equal size for evaluation.
• Income data is adjusted to 2022 dollars and averaged across surveys from ages 30 to 45 for consistency.
• Variables include concrete generational features like log incomes, education categories, race, and sex.

Limitations

• Income data is truncated to an arbitrary range ($5,000 to $448,501.10), which may limit analysis of extreme values.
• The dataset is designed as a class exercise, which may limit its scope and size for production-level research.
• Education is coded from a single report within an age range (30-45), potentially missing lifetime educational attainment.

Provenance

Source

Panel Study of Income Dynamics (PSID) cross-sectional Survey Research Sample.

Collection Method

Data merged across three generations using the PSID Family Identification Mapping System 3-generation prospective linkage file; incomes averaged from several surveys between ages 30-45 and adjusted to 2022 dollars.

Time Range

null

Freshness

null

Geography

null