Eula_Portfolio

Predicting Employment Status Among Foreign-Born Individuals

Skills: Logistic Regression

Census Data Analysis

Feature Interpretation

Classification Modeling

Overview

This project analyzes demographic and socioeconomic predictors of employment among foreign-born individuals in the United States using American Community Survey (ACS) data.

Dataset

The dataset used was the 2024 ACS Public Use Microdata Sample (IPUMS).

The analysis focused on foreign-born individuals aged 18–64.

Key variables included:

• Education level
• English proficiency
• Years in the United States
• Sex
• Age

The outcome variable was employment status during the survey reference week.

Methods

A logistic regression model was used to predict employment status.

Two models were evaluated:

• Baseline logistic regression
• Class-weighted logistic regression to address class imbalance

Model performance was evaluated using:

• Accuracy
• Precision
• Recall
• F1-score

Results

The baseline model achieved strong overall accuracy but performed poorly identifying not-employed individuals. Applying class weights improved recall for the minority class.

The strongest predictors of employment were:

• Education level
• Years in the United States

Tools

Python, pandas, scikit-learn