Data Analyst Scientist
Resume:
Brian Klute – Page *
Brian Klute
Milford, Iowa 51351
**********@*****.***
Analytics leader with 30+ years of work experience executing data driven solutions to increase accuracy of predicted outcomes and provide insights using advanced data driven methods.
• 15 years leading teams in data and statistical analysis
• 15 years using relational databases, SAS, Python, R
• 15 years writing and executing complex SQL queries to extract/process/report data
• 10 years building and validating algorithms
• 5 years with visual analytic tools, Tableau
• 9 years with large, complex data, Python,
Hadoop
Education
Texas A&M University, Houston, TX
Master of Science in Analytics
Summa cum laude
University of Michigan, Ann Arbor, MI
Master of Business Administration, Strategy & Finance Magna cum laude
United States Air Force Academy, Colorado Springs, CO Bachelor of Science in Mechanical Engineering
Cum laude
Skills
Programming Language, Certifications, and Technologies – SAS, Enterprise Miner, SAS Enterprise Guide, Data Robot, Python (Scikit-learn, Pandas, NumPy, Keras, Tensorflow), R, SAS, SQL, VBA, Google Cloud Platform, Tableau, Excel, Business Objects, Access, Hyperion, Monte Carlo simulation, Minitab, Crystal Ball,
@Risk, Microsoft Project, Master Data Management, PowerBI, ETL Tools. Six Sigma Black Belt
SAS Certified Advanced Programmer
SAS Certified Predictive Modeler
Coursera/Stanford Machine Learning Certificate
SAS Certified Base Programmer
SAS Certified Statistical Business Analyst
SAS Certified Clinical Trials Programmer
Udacity Deep Learning Nano Degree
Methods and Techniques – Forecasting; Prophet, Linear/Logistic regression, clustering, nested and segmented modeling, Cox proportional survival regression, generalized estimating equation, generalized linear modeling, resampling, Time Series, Transfer Functions, Decision Trees, Neural Networks, Mixed Models, Kriging, Multidimensional Preference Analysis, Multiple Adaptive Regression Splines, xgBoost, Random Forest, Logistic Regression with stepwise selection, Lasso, Adaptive Lasso, Elastic Net methods, LSTM, CNN, Conditional Logic modeling, Text Mining, Factor analysis experimental design, Support Vector Machines, Extra Tree Classifier, K-means and DBScan Clustering, KNN, Hypothesis Testing, Natural Language Processing
(NLP), Isolation Forest.
Honors, Publications, and Professional Affiliations Texas A&M Analytics Program Full Scholarship
Recipient Tauber Institute Operations Management
Fellow
William Davidson Institute International Business Fellow Member of team that won the 2014
INFORMS Prize for Mayo Clinic.
Dell Premier Performer 20 recipients of
40,000 employees Dell Chairman’s Award
For Quality
Work Experience
Leadership:
Business Growth –
• Led engagement teams to develop new line of business and expanded customer based with 5 key new customers.
• Created proposals that won new business in key new business area.
• Led teams that interacted at client C-Level, including solving emerging issues and reporting on key progress metrics.
Product Development –
Brian Klute – Page 2
• Led process of turning business questions, risks and issues into prototypes, analysis recommendations, test analytics.
• Presented recommendations to key stakeholders at all levels, users and underwriting committee.
• Promoted hypothesis driven approach and rapid prototyping using iterative customer validation. Coaching –
• Mentored junior staff in data wrangling and statistical modeling approaches. Focused on understanding how to translate a business question into valuable predictive analytics techniques.
• Assisted with dimensional reduction and feature engineering to include clustering of principal components and derived classification using similarity metrics.
• Methods developed included time series methods (traditional, ml and deep learning), recommender systems using text classification (deep learning), decision trees and ML for binary classification, nested models for frequency/severity risk assessments.
Data Science Modeling:
High-Cost Event Model –
• Developed event based modeling for high risk youth high cost events.
• Model’s target set custom frequency and severity goals.
• Insights incorporated into daily case management recommendation engine.
• Model development included xgBoost, Random Forest, Logistic Regression with stepwise selection, Lasso, Adaptive Lasso, Elastic Net methods, early stopping.
• Custom Euclidean distance predictor developed using clustering of principal components and distance to event centroid.
Frequency Severity Models –
• Created diabetes and depression segmented and hierarchical model to predict weighted frequency severity of member’s propensity for high cost emergency room and in-patient events.
• Results achieve 6X increase in high risk decile over existing chronic conditions models. Time Series Prediction Model –
• Built demand prediction model that incorporate innovative feature engineering to prepare temporal data features for Prophet, xgBoost and LSTM approach.
• Incorporated linear and cyclical approaches in ensemble to lower forecast standard error.
• Automated model implementation.
Likelihood to Engage Behavioral Model –
• Develop definition of engagement across multiple platforms.
• Used behavioral predictors about lifestyle, demographics, and social determinants of health to create engagement propensity scores.
Cancer Diagnosis Model –
• Featured engineered custom trigger and diagnosis event window for individual member predictions based on a variety of predictors including Rx, diagnosis, conditions, utilization, web interactions, call interactions, including key word and sequential pattern predictors.
• Insights led team to focus on low, medium and high risk outreach strategies. Text Mining Appointment Request Model –
• Using patient’s written text descriptions of symptoms in the appointment request process as input to a long-short term memory neural network model.
• Approach predicted 29 simultaneous practices to assist in defining the patient’s journey in an automated fashion.
• Method increased efficiency and saved labor expense while achieve a 94% overall accuracy level. Calendar Disruption Model –
• Created an event based model to predict cancels, reschedules and no-shows. Methods included machine learning and traditional approaches compared.
• Lead time to cancels/reschedules allows for recovery rates exceeding 85%.
• Created new features for no-show modeling after interviewing physicians.
• New predictor improved accuracy by 55%.
• Insights on key predictors included sweet spot thresholds that management used for capacity planning. Brian Klute – Page 3
Work History
CIGNA/Express Scripts, Milford, IA
Senior Data Scientist
• Built 8 predictive models for stakeholders. Models provided up to 6X lift over existing models at the high risk deciles. Created 2 new models that led to priority based customer engagement strategies to focus on most at risk members.
MAYO CLINIC, Rochester, MN
Principal Data Scientist, Internal Business Consulting Senior Data Scientist, Internal Business
Consulting Data Analyst, Planning Services
• Consult on analytical problems and present solutions options to executive level clients. Provide decision support for business areas across the enterprise.
PROJECT CONSULTING GROUP, Minneapolis, MN
M&A Consultant
• Created requirements for integration associated with Ameriprise acquisition of Bank of American Columbia Investments.
RUST CONSULTING, Minneapolis, MN
Senior Engagement Director
• Managed 5 direct reports and extended staff of 50+
• Created analysis and proposals for current and prospective investment management clients. Engagement at Corporate & Government Board and C-Level. Exposure to External Trustees, Special Masters and chief counsel.
AMERIPRISE FINANCIAL, Minneapolis, MN
• Created comprehensive model to assist in decision making around outsourcing strategy. Approach examined mix changes, people impacts, productivity improvements, costs and benefit timing. Provided real-time decision support during vendor negotiations.
AMERICAN EXPRESS BANK, Salt Lake City, UT
Credit Risk Manager
• Managed 4 direct reports
• Advised AMEX Board Investment & Underwriting Committee on portfolio performance audit and made recommendations for asset allocation, new criteria and parameter adjustments. Developed Markov-Chain model to predict ongoing credit behavior.
DELL COMPUTER, Austin, TX
Strategy Consultant
• Led 4 direct reports
• Completed strategic analysis, developed business cases, conducted due diligence and structured process fix initiatives with $340M EBITDA benefit. Examined cost accounting and analysis of manufacturing process. Top 5 of 23 recommendations accepted and implementations teams created. WHIRLPOOL CORPORATION, Warsaw, Poland, Prague, Czech Republic, Hannover, Germany Strategy Consultant
• Consulted on market entry strategy for central European markets for Joint Venture with German partner. Strategy recommended new price, promotion, positioning and supply chain approach. U.S. AIR FORCE, Cape Canaveral, FL/Dayton, OH
Propulsion Engineer/Program Manager
• Formed joint venture with Industry Canada to implement new technology to an environmental problem for the Air Force. Shepard of idea from concept to launch.
Contact this candidate