Health Data Analysis

Overview

This project performs a full exploratory analysis of the NSMES1988 dataset, which includes demographic, socioeconomic, and healthcare-related attributes for 4,406 individuals. It was developed as part of the Incremental Capstone Session 1, focusing on uncovering insights around hospital visits, insurance coverage, income levels, and health conditions.

Team

• Carllos Watts-Nogueira

• Ranjan Baral

• Course: Artificial Intelligence & Machine Learning

• University of San Diego / Fullstack Academy

• Section: 2504-FTB-CT-AIM-PT

• End Date: May/2025

Dataset Summary

File Name: NSMES1988.csv
Source: Drive Folder
Columns:

• Demographics: age, gender, married, region, school
• Health: health, adl, chronic, hospital, emergency
• Utilization: visits, nvisits, ovisits, novisits
• Economics: income, insurance, medicaid, employed
• Technical Note: Unnamed: 0 renamed to id

Data Summary & EDA (Task A)

** Steps Completed**:

• Renamed columns & cleaned data
• Scaled age × 10, income × 10,000
• Changed age dtype to int16 for memory optimization
• Saved output files: NSMES1988_updated.csv, df2_updated.csv

Key Findings:

• No missing data
• Memory reduced from 654.1 KB → 628.3 KB
• 3 income records show negative values (e.g. -10125)
• Categorical columns analyzed using value_counts()
• Age and income binned for group analysis

Visual Insights & Task B

Insurance vs Income:

• Majority of participants are insured
• High-income individuals more likely to hold private insurance
• Medicaid associated with low-income groups

Hospital Visits by Health Status:

• Poor health correlated with higher physician visits
• Excellent health linked to fewer visits across all categories

Correlation Analysis:

• Heatmaps reveal clear relationships between age/health and hospital visit frequency

Other Insights:

• More females than males in the dataset
• Exploratory plots using jitter, scatter, and bar charts for distribution and relationships

Hospital Data Report (Task C)

This section consolidates all analytical findings:

• Identified usage patterns across hospital visit types by health status
• Found visual trends connecting insurance coverage, Medicaid reliance, and income brackets
• Noted potential biases in gender distribution and health access
• Recommended next steps for anomaly review and deeper cross-variable analysis

Technologies Used

• Python
• Pandas, NumPy
• Matplotlib, Seaborn
• Google Colab / Jupyter Notebook

Next Steps

• Investigate income anomalies
• Build predictive models for insurance or chronic illness
• Visualize age-income-health intersections using interactive dashboards (e.g., Plotly or Streamlit)