Min (Dana) Guan
***** ****** **, ******** *******, CA 91362
310-***-**** ************@*****.*** Professional Experience:
Min Guan is a data-driven professional with strong and extensive experience in statistical analysis and programming. She demonstrated a successful history of developing and implementing new innovative statistical ideas for better business insights. She is a self-motivated learner, and has been proactively seeking new business opportunities for problem-solving and growth.
Work Experience:
J. D. POWER AND ASSOCIATES, Westlake Village, CA
Senior Statistics Manager Jan 2011 – Present
~Led and developed different new advanced statistics modeling approaches to address JDPA business questions and improve existing statistical model performance. Developed brand-new analytical product (rim weighting) to earn new business for JDPA, and developed new Bayesian Network analytical tool and presented it to JDPA research leads and practice leads for new business opportunities.
~Assigned as a peer advisor, mentored new-hire and junior colleagues; Served as a key contact and go-to resource to provide information, answer questions, provide support, and share perspectives in their jobs.
~Led JDPA science team for developing, maintaining JDPA sample weighting processes by designing a series of statistical sample weighting algorithms for JDPA Science team members to apply to different JDPA studies.
~ Led JDPA science team to implement index automation process on remote SAS server by designing a series of SAS automated indexing execution and QA processes. Interacted with JDPA data process team to streamline the automation process and oversee the process on a daily basis. Developed different client deliverable EXCEL VBA statistical tools.
~Provided advanced statistical consulting service based on JDPA survey data from auto retail, auto product, media, health care, insurance, retail banking and some external data for internal and external clients. Provided suggestions on questionnaire design and sampling plan based on JDPA best practice rules. Presented statistical findings to team leads and outside clients.
J. D. POWER AND ASSOCIATES, Westlake Village, CA
Senior Statistician May 2007 – Jan 2011
Responsible for developing JDPA index models for the firm’s syndicated, proprietary, tracking studies from auto, insurance, finance, real estate industries. Provided advanced analytical consulting service for internal and external clients. Established one new statistical process and incorporated it into the existing standard processes. This contribution prevents some extreme respondents from adversely influencing JDPA ranking results. DIVISION OF GENERAL INTERNAL MEDICINE & HEALTH SERVICES RESEARCH, UCLA SCHOOL OF MEDICINE Independent Consultant November 2006 – April 2007
Provided data analysis consulting service to one continuous clinical trial project involved back in school; Co-authored a new paper on that project which focused on identifying the association between change in Estrone Sulfate level and change in mammographic density.
POWER INFORMATION NETWORK, A DIVISION OF J. D. POWER AND ASSOCIATES, Westlake Village, CA Statistician June 2005 – May 2007
Developed statistical models to solve problems related to incentive planning modeling product; Delivered monthly OEM incentive spending reports to outside clients.
DIVISION OF GENERAL INTERNAL MEDICINE & HEALTH SERVICES RESEARCH, UCLA SCHOOL OF MEDICINE Analyst/Programmer March 2005 – June 2005
Participated a clinical trial project and co-authored a paper on that project. Analytical work included analyzing data on several determinants of mammographic density (which is a risk factor for breast cancer) in order to target future breast cancer risk reduction strategies.
NEUROPSYCHIARTRIC INSTITUTE STATISTICS CORE, UCLA
Statistician Assistant Feb. 2004 – March 2005
Provided analytical support on different survey projects conducted through hospital and research centers; Performed database management and data analysis works with SAS macros, and different SAS/Stat, SAS/Base procedures. Select Accomplishments
***Developed Bayesian Network (BN) model on JDPA customer satisfaction data. With BN, several JDPA index key measures can be displayed graphically in a causal network structure. Prediction will not be limited to one outcome. Instead, multiple predictions can be inter-connected to each other, and thus more interesting interpretations can be given. Proactively educated JDPA research team and practice team people on the new BN technique, and used BN modeling to develop new value-add analytical product.
***Developed innovative statistical approaches to solve business problems, address client questions or improve existing statistical modeling performance.
~Proposed and developed different new statistical approaches via R to improve prediction accuracy, such as bootstrapping, cross validation for choosing model tuning parameter, shrinkage methods by ridge/lasso regression, decision tree
~Developed different innovative statistical approaches to perform different survey pilot methodology testing. For example, outlier detection via different regression diagnostic measures; question scale variability measure by unalikeability coefficient estimate; survey question clustering by principal component extraction; Cronbach Coefficient Alpha estimate for measuring internal consistency within the same question cluster,
~Developed new approach for JDPA Key Performance Indicator (KPI or driver) modeling with Regression Tree technique to create improved binary cut point for each KPI. The existing binary cut point selection technique cannot be applied to nominal variable, nor does it take variance into consideration, and is subject to some degrees of arbitrariness when it comes to variables with non-monotonic values. Regression tree approach overcomes these problems, works well with nominal variables, and produces more meaningful binary cuts.
~Proposed and developed Intra Class Correlation (ICC) technique to evaluate how consistent and reliable the results were across different social media analysts in doing sentiment analysis via several purchased social media analytical tools. Applied the same (ICC) technique to Asian studies, and evaluated face-to-face interviewer effect on collecting survey data in Asian market.
~Developed different enhanced clustering approaches to improve market segmentation results for US and China auto markets. For example, applied double standardization instead of single standardization to the clustering variables; generated factor scores/variable cluster scores instead of original scale variables; used clustering ensemble for clustering.
~Developed a new approach of statistical distance calculation (Dot Product) for one JDPA statistical matching project: Data Fusion between JDPA auto online media study and offline Car & Truck media study. The Dot Product distance turned out to be superior to Euclidean/Mahalanobis distance in terms of much larger distance differentiation between matching groups, and therefore, provided better fused results.
~Developed a set of SAS time series prediction processes, and generated predictions for those delayed published AD dollars on a weekly basis.
***Developed one brand-new way of survey sample weighting calculation for JDPA sample raking (or rim weighting). This new technique uses population marginal distribution instead of JDPA traditional joint distribution to weight sample data. Two rim weighting algorithms were developed by using SAS, and had been successfully introduced to JDPA several studies to replace the traditional way of sample weighting practice. In particular, one rim weighting algorithm that can handle empty sampling cell won JDPA a new business from one outside company Compete. With my special rim weighting algorithm, JDPA could move on to collaborate with the company Compete to develop a new clickstream advertising product, and turn it into a monthly deliverable for JDPA. My contribution not only created this new business for JDPA, but also helped bring back about $1.2 million/year revenue to JDPA from this new rim weighting product.
***Developed a series of SAS procedures to automate statistical calculation on SAS server for some standard JDPA tracking studies. The automation was run on a remote SAS server without any human’s intervention. If any abnormal calculation happens based on a set of predefined QA criteria, the SAS server will send out automated email notification to relevant persons for further action.
***Led JDPA science team to develop a set of standard sample weighting procedures to apply to all JDPA studies.
~Developed a set of respondent level sample weight calculation algorithms, and simultaneously screened for outlying sampling cell via Multinomial M statistics.
~Developed enhanced version of sample weighting calculation with the capacity of dynamically adjusting brand level market shares versus their actual sample returns.
~Applied robust regression techniques to detect influential respondent due to unusual survey response pattern and/or extreme sample weight, and evaluate its adverse impact on regression weights and final scores.
***Developed Excel VBA tools which become standard deliverables used by the team members and clients.
~Developed Multinomial M statistics for detecting outlying sampling cell and developed several versions of Excel sample weight screening tools with embedded screening algorithms. The tool is used for screening for outlying sampling cell which may cause adverse impact on final satisfaction score.
~Developed Excel Key Performance Indicator (KPI) simulator. The tool becomes an easy template for team members and clients to use, and greatly facilitates JDPA sales team to present JDPA index model to outside clients. The tool simulates how JDPA index score performs relative to its KPIs’ performances based on the regression between JDPA index score and its set of KPIs.
***Led the JDPA science team to build a SAS data warehouse through advanced PROC SQL VIEW features with the aid of the SAS macro facility and SAS runtime dictionary tables to demonstrate how this data warehouse system can allow user to perform data mining from different industries across time in one centralized location. Relevant Education Background:
M.S. in Biostatistics (June 2005)
UNIVERSITY OF CALIFORNIA – LOS ANGELES
SUN MICROSYSTEM Certified Programmer for Java 2 Platform (March 2001)
Computer Science Certificate on Database Technologies (Nov. 2000) UNIVERSITY OF CALIFORNIA – IRVINE
Computer Languages & Skills:
SAS, SPSS, R, VBA, SQL Stored Procedure, PL/SQL, Java, C++
Microsoft 2010 Excel, Access, Power Point, Word
Relational Database, Object Oriented Programming
SQL Database Sever 2008, Oracle Database 11g
Reference: Available on Request
Professional Membership: American Statistical Association (ASA) Publication:
Carolyn J. Crandall, Min Guan, Gail A. Laughlin, Giske A. Ursin, Frank Z. Stanczyk, Sue A. Ingles, Elizabeth Barrett- Connor, and Gail A. Greendale. Increases in Serum Estrone Sulfate Level Are Associated with Increased Mammographic Density during Menopausal Hormone Therapy. Journal of Cancer Epidemiol Biomarkers Prev 2008;17(7):1674–81. Carolyn J. Crandall, Arun Karlamangla, Mei-Huan Huang, Giske Ursin, Min Guan, Gail A. Greendale. Breast Discomfort During Hormone Therapy Predicts an Increase in Mammographic Density. Archives of Internal Medicine 2006;166:1578- 1584.
David A.Ganz, John T. Chang, Carol P. Roth, Min Guan, Caren J. Kamberg, Fang Niu, David B. Reuben, Paul G. Shekelle, Neil S. Wenger, Catherine H. MacLean. Quality of Osteoarthritis Care for Community-Dwelling Older Adults. Journal of Arthritis & Rheumatism (Arthritis Care & Research) Vol. 55, No. 2, April 15, 2006, pp 241–247.