Dahai Huang

EV Sales Forecasting

• Built predictive models (logistic regression, decision tree, Naïve Bayes) in Python to forecast new energy vehicle adoption.
• Achieved 92% accuracy; evaluated model performance via ROC-AUC and confusion matrices.
• Provided insights on marketing strategy and future demand forecasting to support sales growth.

Bitcoin Network Security Analysis

• Modeled Bitcoin network stability using R with 450+ daily observations of hash rate, difficulty, transaction value, and etc.
• Detected autocorrelation in residuals and applied Cochrane-Orcutt estimation to correct errors, improving model reliability.
• Addressed multicollinearity with Partial Least Squares regression, achieving predictive R² of ~80% (training) and 84.5% (test), confirming strong predictive power.
• Found that miners’ activity (hash rate + difficulty) emerged as a key latent driver of resilience, offering insights into how volatility impacts Bitcoin network stability.