Problem Description

• The U.S. Census Bureau’s Annual Business Survey (ABS) dataset provides valuable insights into business and demographic statistics across various industries.

• Our project aims to leverage this data to explore trends and characteristics of employer firms, with a focus on factors influencing business growth and workforce composition.

• Goal: This analysis can aid policymakers, researchers, and businesses in making informed, data-driven decisions and our goal is to ensure our analyses yield meaningful and reliable insights into business and workforce trends within the Census dataset.

Why is this interesting?

Through this project, we intend to uncover:

• Industry growth patterns
• Demographic disparities in business ownership
• Workforce distribution across sectors

Key Analyses Plan and Insights

• Identifying Key Predictors of Employer Firm Counts:
  • Using demographic attributes, business characteristics, and geographical data, we will analyze factors significantly impacting the estimated number of employer firms (OWNPDEMP).
• Comparative Analysis by Demographics and Industry:
  • Analyze trends by gender, ethnicity, veteran status, and industry classifications to highlight disparities or growth patterns, offering insights into specific business sectors and demographics.
• Temporal and Regional Trends:
  • Utilize YEAR and GEO_ID data to study changes over time and by region, identifying areas with high growth or potential support needs.

Evaluation:

We will evaluate our results through a combination of model performance metrics, clustering quality, and feature importance analysis. Specifically, we’ll use RMSE and MAE to assess prediction accuracy, silhouette scores to gauge clustering effectiveness, and consistency checks across feature importance rankings to ensure robust insights. Additionally, we’ll analyze trends by demographics, industry, and region to validate the practical relevance of our findings.

Models We will be using for final results:

• Decision Tree

• Random Forest

• Keras Library for clustering

• Gender based employees distribution among industries.

Data Visualizations:

• Top Industries by Average Number of Employees: A bar chart illustrating industries with the highest average employee counts.
• Employee Distribution by Ethnicity: A density plot showing employee count distributions across different ethnic groups.
• High-Growth vs. Low-Growth Industries: Comparison of average employee counts between high-growth and low-growth industries.
• Mean-Median Difference (Skewness): Highlight industries with significant skew in employee counts.

Tools & Libraries: We will utilize R packages, including ggplot2 for visualizations, dplyr for data manipulation, and relevant machine learning libraries for model development.