Developed an econometric pipeline to detect fraud risk in 78,600 metaverse transactions using OLS regression, PCA, and HC3 robust standard errors. Identified veteran cohorts with twice the risk and phishing as the strongest fraud signal.
From the README
Metaverse Cyberfraud Shielding Econometric Risk Detection Python Project Overview Built an econometric pipeline to detect and quantify transactional fraud risk across 78,600 metaverse transactions. OLS regression was used for risk score modelling; PCA for multicollinearity correction; and HC3 robust standard errors for valid inference under heteroskedasticity. Tools & Technologies • Python — end-to-end pipeline scripting • statsmodels — OLS regression, White's heteroskedasticity test, Durbin-Watson test • scikit-learn — PCA, StandardScaler, OneHotEncoder, train-test split • scipy.stats — Shapiro-Wilk and Kolmogorov-Smirnov normality tests • seaborn / matplotlib — EDA visualisations and...
Skills
- Jupyter Notebook
