SUMMARY

• Extensive industry analytics experience using Predictive modeling and Data Mining.L.L. Survila has over 15 years Clinical data analysis experience and 8 years of Marketing and Sales data analysis experience, has been a Predictive Modeler, Data Miner, Researcher.
• Experienced Predictive Modeler, Data Miner skilled in gathering and documenting requirements, developing project plans, creating business process models and flows, managing and tracking project execution, developing and executing test scenarios.
• Comprehensive experience in analyzing end user requirements and problem solving skills with strong analytical background, ability to define problems, collect, clean and discover hidden value in data, draw valid conclusions, and present results.
• Capable to develop, validate predictive models of rare and uncommon cases e.g. rare disease, rare genes mutations, intrusion, medical, credit card or insurance risk or frauds analysis.
• Working knowledge of in applying a wide range of Data Mining, Machine Learning, Statistical Analysis and Optimization methods like Neural Networks, Decision trees, Association rules, Cluster analysis, Support Vector Machines, Bayesian Networks, Classifier ensembles, Linear Programming, Response Surface methods, Genetic algorithms, etc. to Clinical and Marketing data.
• Proficient in predictive data analysis such as predictive feature extraction, predictive modeling of highly imbalanced data, model aggregation / fusion, boosting, bagging, linear/nonlinear regression, classification, segmentation, profiling and hypothesis testing with the tasks, duties and responsibilities focusing on identify salient data trends and patterns, create, validate and maintain predictive data models, document and present the result.
• Skilled to analyze customer & prospect data to gain deep understanding of customer segment and behaviors.
• Ability to create Market Segmentation (Behavioral, Value - Based, Propensity-Based, Needs/Attitudinal-Based, Loyalty, Social-demographic), Response, Risk (potential for loss), Net Present Value (product overall profitability), Cross-Sell and Up-Sell, Price Elasticity models.
• Good SAS programming skills (SAS/Base, Proc SQL, SAS/STAT, SAS/Macro, SAS/Access, SAS/ODS, SAS Graph) acquired in 6 years experience in analysis, design, development, testing and implementation of various SAS applications.
• Experienced to develop and execute complex, advanced SQL queries and select from multiple data sources and capable of mapping and consolidating data into a single list.
• Skilled to collaborate with the mid-senior business, research, medical & marketing staff to complete joint reports.
• Accomplish objectives and work well without direction. Able to identify and learn new technical skills.

TECHNICAL SKILLS

Statistical: Data mining toolsSAS, SAS Enterprise Miner, SPSS, SPSS Modeler (Clementine), SPSS Answer Tree, R, CART, CHAID, DTREG, Weka, Tiberius

Programming languages: SAS, SQL, VBA, FORTRAN, Python, HTML, XML, UML

Operating Systems: Windows, Linux, DOS

Personal productivity tools: Excel, PowerPoint, Word, Visio

PROFESSIONAL EXPERIENCE

Confidential, Englewood, CO

Data Mining Architect/ Analyst

Responsibilities:

• Being the part of Advanced Analytics team brought scientific rigour to direct marketing efforts, leveraging customers and prospects data to build robust & accurate predictive models of customer behaviour.
• Performed data mining on customers and prospects data bases.
• Built True Lift (Incremental Response model), studied scientific literature on the matter, used balancing and model aggregating techniques to stabilize classifiers. (SAS EM)
• Built carrier route model for direct mailing campaigns.
• Built DISH vs DirecTV propensity model.
• Significantly improved accuracy and reliability of previously built response model.
• Worked closely with business senior intelligence.
• Manipulated large transactional databases (250 millions records) using Teradata SQL and SAS programming.

Confidential, Auburn Hills, MI

Predictive Modeler / Project Manager

Responsibilities:

• Evaluated business objectives, assessed situation, determined data mining goals, produced data mining project plan for service contract (extended warranty) project.
• Designed framework to predict service contract cost based on quality of vehicle subsystems, wear out, usage pattern, geographic location (temperature, humidity, altitude, urban/ suburban/ rural, labor cost) and optimize service contract price. Investigated and selected relevant modeling techniques (SVM, Bayesian networks, ANN, Decision trees). Combined selected models to ensemble (SPSS IBM Modeler).
• Proposed to use Text mining to identify patterns in the text and predict outcomes that describe product quality problems.
• Assumed leadership role in the utilizing of data-mining techniques/capability within the program.
• Developed executive-level presentations to higher business management.
• Coordinated team pursuits in alignment with project objectives. Collaborated with the offshore team.
• Communicated analytical findings to engineering and/or business analysts on a regular basis.

Confidential, Lansing, IL

Predictive Modeling

Responsibilities:

• Developed management decision support models and reports. (R)
• Performed data mining analysis to predict periodic customer requirements.
• Performed forecasting and customer demand planning (Regression analysis, Decision trees, ANN, Bayesian networks). (R)
• Developed load-matching model.
• Developed short-term operations planning model.
• Proposed operational model for vehicle dynamic routing and scheduling.
• Performed facility location analysis and total costs throughout company minimization.
• Created summary reports, conducted statistical analysis of key business performance indicators (Status of existing customer, Customer attrition (CHAID, CART, decision tree forest, boosted decision trees, logistic regression), Segmentation of customers by profitability, Turnover generated by segments of the customers, etc.) (R).
• Specified and discussed intelligence and information requirements to design Decision Support System with internal and external staff.

Confidential, Hinsdale, IL

Data Mining Analyst

Responsibilities:

• Performed quality data analysis, cleansing, presentation, reports using relevant tools and techniques.
• Evaluated the business data to reveal the profitable and non-profitable areas of revenue, made decisions about the cost incurred on different activities.
• Developed an algorithm to match client with various types of diseases with care worker based on number of tests using multidimensional similarity measure, selected set of influential indicators. Tested various distance metrics like Euclidean, Mahalanobis, Manhattan, Angle Between Two Vectors, Number of Features in Common.
• Studied advertizing efficiency, made recommendations.
• Developed client/ customer data base and supporting applications.

Confidential

Data Mining Analyst / Predictive Modeler / Researcher / SAS Programmer

Responsibilities:

• Modernized an IT environment from legacy systems to newer database, created mapping specifications for the movement and transformation of data from source to target, performed data cleansing.
• Performed Quality control for data migration from an old database model to a new model using SAS.
• SAS, R, CART, DTREG statistical software and homemade data mining tools, neural networks, logistic regression, decision trees, association rules, Bayesian networks, model ensembles based on boosting and bagging, attribute importance, error-correcting output codes, table analysis, and optimization methods were used to categorize data, investigate patterns and look for significant trends.
• Studied data mining techniques like decision trees C4.5, Self-Organizing Maps, Support Vector Machines, Neural networks, under sampling and oversampling techniques, classifier combining methodologies to analyze highly imbalanced data sets, elaborated algorithms suitable for uncommon cases prediction e.g. rare disease, rare genes mutations, intrusion, medical, credit card or insurance risk or frauds analysis. Performed qualitative comparison of algorithms by effectiveness, scalability and speed.
• Developed model to predict In-Hospital modes of death after reperfusion therapy for ST elevation Myocardial Infarction. (R)
• Studied selection of the optimal non exercise stress for the evaluation of ischemic regional myocardial dysfunction and malperfusion, designed and coded SAS programs to implement constrained optimization. (SAS)
• Performed data extraction from Oracle, Access databases & Excel using SAS/SQL.
• Performed SAS macro code modification for reports and tables.
• Developed model based on decision trees ensemble to evaluate short term mortality risk in Acute Myocardial Infarction. (C4.5, C5.0, CHAID, SAS)
• Elaborated Noninvasive Hemodynamic Index for immediate prognosis after Acute Myocardial Infarction. Compared different groups of patients according this index (Anova, SAS, Access).
• Coded program for evaluation of the coronary circulation (Java, SQL).
• Developed model for quantitative assessment of Acute Myocardial Infarction size. (SAS)
• Elaborated several left ventricle hemodynamic and contraction models based on the 12-lead ECG. (Java, SAS).
• Created macros using Visual Basic for Applications (VBA) for Excel spreadsheet to perform mathematical calculations and predict patients (Logistic regression) outcome after angioplasty.
• Analyzed clinical data and generated various statistical reports using SAS, SPSS, R, CART.

Confidential, KAUNAS, LITHUANIA

Predictive Business Data Analyst

Responsibilities:

• Created Response, Risk (potential for loss), Net Present Value (product overall profitability), Cross-Sell and Up-Sell, Price Elasticity models.
• Performed Market Segmentation Analysis, used logistic regression, decision trees, cluster, principal components and factor analysis to combine attitudinal and demographic and other data to develop Behavioral, Value-Based, Propensity-Based, Needs/Attitudinal-Based, Loyalty, Social-demographic and Life-Stage segments that are easier to target.
• Developed and improved models and processes to consolidate and analyze the annual operating plan, strategic plan and quarterly forecasts.
• Collected information about customers submitted through the registration forms and generated by the web-log analyzing applications, performed data cleansing. Used quantitative analysis technique to analyze this information and produced marketing reports to identify potential customer segments.
• Created user accounts and individual marketing targets for users.
• Constructed various custom, interactive, on-line tools to facilitate the marketing process, such as - financial calculators, price forms, autoresponders and mailing lists.
• Developed a sales information processing and analyzing system, including order processing, inventory control, materials expenditure, received/paid accounts.

Confidential, LITHUANIA

Research Fellow/ Data Analyst.

Responsibilities:

• Leaded software part of the project to develop a computerized system for non-invasive evaluation of intracardial and peripheral hemodynamic and left ventricle regional and global functional injury. Responsible for project planning, determining deliverables, and estimating time frames. Standards were established for programming, testing, and validation.
• Developed Predictive Clinical Score Index for severe morbidity after coronary artery surgery.
• Implemented Preoperative Clinical Severity Score Index for coronary artery bypass patients (SAS).
• Coded interactive search tools to query and repair the database; programmed data classification, presentation and statistical analysis modules.
• Coded data input, cleansing and verification modules (SAS, SQL).
• Studied robust discriminant functions. Designed and clinically tested a number of robust discriminant analysis functions for the MI outcome prediction.
• Developed data base formatting and transporting tools between different architecture computers: from HP9630 to IBM PC and IBM mainframe.
• Developed statistical analysis subsystem DBSAS with 28 statistical procedures for the HP DBMS Image/1000. Data search module exceeded similar Hewlett-Packard module from the Image/1000 Query reporting system twice in a search speed (regular grammar theory and assembler programming language were used).