Statistical Analysis for Business Insights - Regression and Time Series Modeling
This project demonstrates my advanced statistical modeling skills applied to real-world business questions, highlighting my readiness to drive data-driven decision making as a future data scientist.
Project Overview
- Regression Analysis: Studied the impact of tax exemptions on fertility rates using causal inference models in Stata.
- Time Series Analysis: Explored Airbnb rental trends on Lake Erie, uncovering seasonal cycles and links to retail clothing sales patterns.
- Model Diagnostics: Addressed serial correlation with Durbin-Watson testing and corrected for heteroskedasticity with robust standard errors.
Detailed Workflow
- Exploratory Data Analysis (EDA) to understand distributions, outliers, and missing patterns.
- Regression modeling with multiple independent variables, including interaction terms where relevant.
- Durbin-Watson Test to detect and quantify autocorrelation in residuals.
- Application of Huber-White heteroskedasticity-consistent standard errors for more reliable inference.
- Seasonality decomposition of Airbnb rental time series using additive models.
- Visualization of trends, seasonality, and residuals for clearer business interpretation.
Statistical Techniques Applied
- Multiple Linear Regression
- Time Series Decomposition (Trend + Seasonality)
- Durbin-Watson Autocorrelation Testing
- Robust Standard Errors (Heteroskedasticity correction)
- Model Interpretation and Business Communication
Business Impact
- Quantified the relationship between tax policy changes and demographic outcomes (fertility rates).
- Enabled seasonal planning for Airbnb operators based on rental demand cycles.
- Supported clothing retailers by forecasting seasonal shifts connected to vacation and rental activities.
Reflection
This project strengthened my ability to not only build statistical models but also validate them rigorously and translate quantitative findings into strategic business recommendations. It demonstrates my readiness to apply econometric, regression, and time series analysis at a professional data science level.