Amanda Flaxman Email & Phone Number

Developed a robust document analysis framework integrating OCR (Tesseract), image processing (PyMuPDF, OpenCV), NLP techniques (spaCy), with comprehensive logging for enhanced traceability and debugging. Overcame non-standard PDF encodings to improve data extraction reliability and significantly improving operational efficiency.

Persona Segmentation Leadership: Directed task force within marketing data science team to conceptualize and deploy a data-driven advertising program. Influenced targeted advertising for 230M Americans by identifying 24 unique characteristic and spending based persona segments.Revenue Generation: Agency secured over $10M in new business revenue using innovative persona-based advertising methodology developed by my team.Predictive Market Analysis: Increased luxury brand retention, name recognition, and improved audience development using cluster analysis and psychographic predictive methodologiesClient Engagement: Collaborated with internal client-facing teams to relay key findings, created sales and educational decks effectively communicating complex analysis to external clients.Innovative Process Implementation: Championed DataBricks adoption for enhanced Python execution. Improved complex data engineering tasks e.g. data encoding, dimensionality reduction via principal components, and hierarchical clustering for nuanced categorization.Team Upskilling & Training: Educated data science team on both advanced analytical models and novel analyses. Ensured team can handle future requests, produce refined outputs, and provide audience behavior summaries effectively.Professional Development: Expanded Python/Pyspark, DataBricks, and Snowflake/SQL integration expertise. Provided focused training to task force members on advanced analytical methods, including principal components analysis and hierarchical clustering.

Automated Reporting: Developed automated reports leveraging newly-acquired marketing data assets, detailing audience viewing habits to optimize targeted advertising strategies.Client-Centric Impact: Established a new reporting standard catering to the needs of the client-facing data solutions team, offering deep insights into audience preferences for television consumption, purchasing habits, and demographics.Cross-Functional Collaboration: Actively collaborated with the client-facing data solutions and Media Technology & Data (MTD) teams to ensure streamlined data processes and effective information dissemination.Tech Stack Proficiency: Mastered the utilization of R, Python, and SQL/Snowflake to handle complex data processing and analysis tasks.Leadership & Training: Spearheaded 'lunch and learn' sessions and guided a rapidly-growing team of 10 to 20+ members on best practices and process adoption.

Chicago, Illinois

- Create dynamic web applications using R Shiny, Azure DevOps, and Git to automate specific processes and convey insightful results via interactive data visualization - Predict desirable MSAs to live in amidst coronavirus pandemic. Use k-means clustering algorithm to classify growing, declining, and stable areas based on real estate demand, then perform a multinomial LASSO regression for 10GB of data containing potential migration drivers. Dimensionality was reduced significantly, and prediction accuracy was above 90%. Each MSA was scored on an index 0-100 to reflect the probability of someone moving to that MSA- Construct multiple imputation using random forest approach for hospital service costs, increasing accuracy of results of the final analysis compared to imputing the mean of those costs- Build cluster analysis for a mobile veterinary consolidator to understand their consumer base for impactful targeted advertising and optimize veterinary route scheduling - Generate logistic regression to calculate the probability of customers for a mobile veterinary consolidator existing within a geographical area as well as calculating their addressable market - Increase efficiency within financial due diligence analysis by identifying the optimal maturation month for a new clinic when creating pro forma financial schedules, which allows clients to forecast their revenue, EBITDA, and volume of foot traffic. Integrate R within Ateryx to automate this process and display data visualizations and other insights within Tableau - Identify opportunities for utilizing third-party data as well as machine learning applications by isolating potential business end-use cases among different subsectors of healthcare and life sciences- Perform exploratory data analysis, standard financial reconciliation, and query/clean data for financial and statistical analyses

Chicago, Illinois

- Data science/informatics analyst duties include providing statistical support, managing data, and incorporating clinical knowledge into analysis and reporting. Helps with collecting, cleaning, and analyzing data at all stages of research/development, implementation, and analysis. Completes projects using multiple software programs, such as R, SAS, Python, and SQL (Vertica). Phrases findings in an easily interpretable format, such as in an Excel spreadsheet or PowerPoint presentation, and explains results at the top-level.- Generated a Bayesian logit model with random effects that specifies by point-differential to be used in a ranking algorithm, which rates specialists based on health index scores (developed in-house) of BCBS members with relevant complications. Incorporated cluster analyses specifying each member’s health index status magnitude and trend over a course of five years.- Responsible for the unsupervised analysis addressing high cost claimants for BCBS members. Optimized the process flow of how ~21,000 members exceeding $250K in claims with ~4,000 factors fit into latent groups. This grouping is done in an effort to implement targeted interventions preventing members from reaching very high costs via utilizing predictive maintenance methodologies and recursive neural networks.- Applied knowledge of Bayesian statistics when calculating minimum sample size necessary from average length and coverage criterion and worst outcome criterion calculations. Applied similar methodology for frequentist Wilson Interval minimum sample size calculations. Created functions to run these calculations with different parameters that the calculation calls for (such as interval length distance or specificity).- Mined association rules for diagnosed conditions in Medicare patients- Performed elastic net and non-negative linear regression calculating positive risk scores in Python

Ravenswood, Il

- Find key factors (i.e. if members are connected with a primary care physician, etc.) that could potentially reduce hospitalization costs and rates- Generate business insights that will determine which risk factors are associated with high-user (high hospitalization & ER rates) status - Address high-user members and create a multivariate model to appoint which risk factors are significant, identify any interactions, and predict which factors could increase risk- Perform random complete block design tests to detect whether having a primary care physician reduces the number of hospitalizations across FY14, FY15, and FY16

Arlington Heights, Il

- Designed fraction factorial block experiments testing different anti-corrosive paints (i.e. graphite-infused, commercial standard, etc) in a saline spray machine in order to find the optimal paint that will withstand conditions on an oceanic coast- Calculated MANOVA and repeated measures of results for experiments involving coated/uncoated steel panels exposed to accelerated weathering, galvanic, and saline spray machine testers- Generated response surfaces to find the best set of factor levels to identify the highest-withstanding steel paint coating- Created a front- and back-end web application for battery cycling data in RStudio (Shiny application)- Used statistical analysis and construct mathematical algorithms to guarantee correctly programmed machinery and a high-quality product output

- Framed business questions that allowed for the investigation of serious, transmission, and distribution events based on behavioral audit data in an effort to prevent infrequent incidents- Queried behavioral audit and incident event data from an Oracle Hyperion database to perform multiple statistical analyses in R and proposed important changes to upper management based on results- Applied exploratory analysis, factor and principal components analysis, machine learning principles, variable selection algorithms such as Lasso regression and AIC to multivariate data, and created a unique nonparametric time series forecasting algorithm with exponential smoothing - Designed a detailed procedure instructing all members of the Performance Assessment Group how to perform certain statistical algorithms, why they are necessary, and how the results can impact how business is conducted in the future

- Study linguistic rhythmic differences between American English to British English and other types of English in written texts- Encode context tree weighted algorithms (variable length Markov chain) (binary sequential procedure, created in R) to display the subtle differences in dialects of English based on syllable stress, phonological words, and intonation in texts- Calculate the Bayesian Information Criterion context tree estimator to create champion trees (maximizes likelihood)

Arlington Heights, Il

- Conduct innovative research and testing on new graphite and lithium battery products to ensure energy efficiency and longevity- Analyze properties of graphite under nanotechnological examination- Utilize data analysis and construct mathematical algorithms to guarantee correctly programmed machinery and a high quality product output- Oversee battery production including electrode creation, assembly, and glove chamber workSummer Intern Program Coordinator- Train and delegate tasks to all of the summer laboratory interns in AETC's STEM development program- Manage intern progress and development

Arlington Heights, Il

- Assisted laboratory engineers with ensuring energy efficiency in battery products- Developed laboratory acumen through first assisting and then conducting research in atomic absorption spectroscopy, cathode production, and properties of graphite- Provided support with data analysis and project math model building

- Assisted with the recruiting, hiring, and training of new student callers- Monitored, coached, and supervised calls to alumni, parents, and other constituents- Developed competency using RuffaloCody’s CampusCall software- Managed and produced data analysis reports detailing participant characteristics and vital information- Overall stats – Dollars: $12,221.12 Pledges: 85 Credit cards: 26

Master Of Science (M.S.), Statistics

Bachelor Of Arts (B.A.), Mathematics; Environmental Studies

Activities and Societies: League of Environmental Awareness and Protection, Delta Delta Delta, Four-Wall HandballGraduated with double.