Post Job Free

Resume

Sign in

Statistical Analysis

Location:
Los Angeles, CA
Posted:
February 08, 2018

Contact this candidate

Resume:

Xiaoyang ZHOU

*** ****** ***, *** **, Los Angeles, CA, 90038

951-***-**** • ac4ego@r.postjobfree.com

EDUCATION

Ph.D. of Applied Statistics, University of California, Riverside Dec 2017 Master of Statistics, University of California, Riverside Jun 2014 Bachelor of Economics, Zhengzhou University, China Jun 2012 EXPERIENCE

MUFG Union Bank, San Francisco, CA

Credit Risk Analyst Internship

Jun 2016 – Sep 2016

• Constructed Logistic Regression model to forecast Probability of Default (PD) with low failure rate longitudinal dataset.

• Applied classification method, ROC Curve (Receiver Operating Characteristic), to evaluate model/factor’s predictive power by comparing Area Under Curve (AUC).

• Modified the model of Allowance of Credit Loss (ACL) by integrating three statistical models

(LGD/EAD/PD models). Validated the performance of modified ACL distribution with Monte Carlo simulation which demonstrated an improvement in Actual Loss estimation. Department of Statistics, UC Riverside

Research Assistant

Sep 2014 – Dec 2017

Independent Component Analysis (ICA) project

• Developed an innovative machine learning algorithm of Independent Component Analysis (ICA) algorithm upon separating mixed signals based on Fisher Discriminant Information Matrix (FDIM), named FDIM-ICA.

• Estimated unknown density in FDIM-ICA using nonparametric (Kernel Density Estimation) and semi-parametric (Gaussian Mixture Model) methods.

• Validated the performance of FDIM-ICA algorithm by simulation study and applied FDIM-ICA algorithm to detect and eliminate the white noise in mixed signals.

• Developed an unsupervised clustering method based on FDIM-ICA algorithm and confirmed its superior performance by comparing with PCA/Bayes clustering in real data application. Electric Load Forecasting project

• Collaborated with Electrical and Computer Engineering Department, UC Riverside on forecasting electric load in Southern California Edison’s service territory.

• Cleaned, aggregated raw household level longitudinal dataset and proposed a robust mixed-effects segmented regression model to investigate the significant influential factors relative to electric load.

• Developed a back-fitting algorithm using computational optimization methods (EM algorithm and Newton-Raphson algorithm) to estimate model coefficients.

• Applied stratified sampling method to select predictive dataset and demonstrated the superior performance on prediction accuracy by comparing with three other published models. Xiaoyang ZHOU

Department of Statistics, UC Riverside

Statistical Consultant

Sep 2013 – Jun 2014

Insect growth rate (IGR) vs. temperature study for Entomology Dept. of UC. Riverside

• Applied non-linear model to study IGR under stable and fluctuating temperature; estimated parameters with Gauss-Newton and Newton-Raphson algorithms and calculated variance with Delta method in R.

Insect control on grapes study for Entomology Dept. of UC. Davis

• Constructed generalized linear mixed model (GLMM) to analyze potential relationships between grape disease and grape mealy bugs; applied pseudo-likelihood algorithm to estimate parameters and understand sources of variation by SAS. Examined the GLMM model results with decision tree. Monkey immune system research for Medical Center of UC. Riverside

• Established time series model to assess immune responses of monkeys and calculated the infection level with AUC method.

Department of Statistics, UC Riverside

Teaching Assistant

Sep 2014 – Jun 2017

• Guided labs and led discussions for graduate and undergraduate statistical courses.

• Provided guidance on statistical software such as R, SAS, Minitab and Excel.

• Received Outstanding Teacher Assistant (OTA) Award in UC Riverside 2016. Course taken at graduate level Sep 2013 – Jun 2015

• Advanced Theory of Probability and Statistics, Bayesian Statistics, Nonparametric Methods, Computational Statistics, Categorical Data Analysis, Sampling Theory, Statistical Consulting and Data Analysis, Multivariate Analysis, Advanced Design and Analysis of Experiments. SOFTWARE & LANGUAGE

• Fluent with R, SAS (SAS Certified Base Programmer for SAS 9)

• Experience with Python, SQL, MATLAB, and computing in Unix/Linux environment. PUBLICATION

• Zhou, X., Yu, N., Yao W. and Johnson, R. (2015) Forecast Load Impact From Demand Response Resources. 2016 IEEE PES General Meeting. (Best Paper Prize in IEEE Conference)

• Zhou, X., Yao W. (2017) Fisher Discriminant Information Matrix and Its Application to Independent Component Analysis. Preprint submitted to Journal of Computational and Graphical Statistics. (Under peer review)

• Zhou, X., Yu, N., Yao W. (2017) A Robust Mixed Effects Segmented Regression Model for Electric Load Forecasting. Preprint submitted to Annals of Applied Statistics. (Under peer review) REFERENCE

• References available upon request



Contact this candidate