Yang Deng Email & Phone Number

Pittsburgh, Pa, Us

Selected courses:Operating Systems, Introduction to Algorithms, Introduction to Data Structures, Machine Learning

Columbia, Md, Us

Customer Response Rate Predictions for Banks• Led data analysis for Fifth Third Bank and KeyBank to select customers with high response rates to mortgage offer letters using logistic regression.• Implemented a parallel processing framework with SAS Macro to reduce data ingestion time by 80%.• Conducted feature selection from more than 2000 variables, and evaluated model using Hosmer-Lemeshow test, ROC curve and confusion matrix.• Analyzed the actual customer responses and made reports in the following campaign.

Pittsburgh, Pennsylvania, Us

Brain Image Research• Studied relationship between income and brain image using ordinary least square regression (OLS).• Proposed regression regularization to achieve clearer interpretation and higher overall prediction accuracy.• Shrunk more than 50% coefficients by lasso regression, and significantly reduced mean squared error on the test set.

Pittsburgh, Pa, Us

Non-line-of-sight (NLOS) Global Navigation Satellite System (GNSS) signal Classification• Co-implemented classifiers using various models including logistic regression, decision tree, nave Bayes, and support vector machine to check the existence of NLOS GNSS signals for in-car navigation applications.• Processed highly imbalanced data using techniques such as bootstrap resampling.• Evaluated the performance of the trained classifiers with the area under ROC curve and overall accuracy.Ozone volume Forecasting• Proposed an ARIMA approach for real-time air pollution prediction, and used historical ozone volume data from an air quality station for evaluation.• Determined AR and MA factors using autocorrelation function (ACF) and Partial ACF, and selected the final model using residual diagnoses, Akaike information criterion and Bayesian information criterion.• The prediction based on the final ARIMA model fitted well on the test set, with small mean square error.

Hysterectomy Kits Waste Discharge Comparison• Applied ANOVA test to compare the waste discharge from different types of hysterectomy kits.• Investigated departures from normaility using Q-Q plots and applied Brown-Forsythe test for variance constancy.

M.A., Statistics

Cross Registration, Machine Learning

B.A., English Language And Literature, General