Data Scientist
2021 - 2021The Spur Group- Built binary and multi-class NLP models to determine violations for vitamin supplement reviews.
- Explored the entire history of available data to select the most appropriate training and validation data sets.
- Uncovered data issues relating to change in business practices and explained how this relates to model performance.
Technologies: Natural Language Processing (NLP), KerasPrincipal Data Scientist
2014 - 2021Data Science Evolution- Modeled compensation distributions (5th, 25th, median, 75th, 95th percentiles) using Bayesian Networks on limited data fields, exceeding the performance of a much more complicated champion model.
- Built second-party fraud models in the automotive lending space on behalf of lenders, uncovering bad dealerships and other actors.
- Used health insurance claims data to understand the effects of new pharmaceuticals in the pharmaceuticals marketing space.
- Created semi-supervised segmentation for health care professionals based on their propensity to prescribe newer treatments.
Technologies: Bayesian Inference & Modeling, Data Mining, Jupyter Notebook, Python, SQLLead Data Scientist
2016 - 2018SteppeChange- Built a series of solutions for customer personalization in the telecommunications space for a large EU telecommunications client, serving 20 million customers.
- Led several parallel use cases in course of the data management platform project with terabytes of daily data.
- Created models to determine the probability of a customer exceeding their voice, text, and data allotments. Incremental spend expectation once this threshold was exceeded.
- Analyzed whether a customer was rationing their usage to manage their allotted thresholds.
- Used the outputs of the models and other data to create the next best offer recommendations and pricing, keeping in mind the overall impact on the business.
Technologies: Python, Apache Hive, Spark SQL, Customer Segmentation, CatBoost, XGBoost, Scikit-learnVP, Data Science
2011 - 2014LexisNexis Risk Solutions- Built and led a team of highly skilled data scientists and engineers to solve challenging problems for well-known financial, retail, and wireless institutions.
- Collaborated with the product organization to develop next generation credit risk and fraud models and attributes, seeking to derive incremental and unique value from internal data sources combined effectively with third-party data.
- Advanced research and development efforts, including cutting-edge custom modeling algorithm development, creation and adaptation of new data sources, and developing new verticals.
- Created a proof-of-concept to uncover Federal and State tax identity fraud to combat a multimillion-dollar problem where bad actors were using other people's personal identities to get returns sent to them.
- Calculated the Customer Lifetime Value and built attrition models on LifeLock customers.
Technologies: Fraud Prevention, Credit Risk, Management, Leadership, Mentorship & Coaching, Team Building, Executive PresentationsDirector, Analytics
2006 - 2010First Data- Led custom analysis in the financial services space, building client-specific models and taking the problem from raw data to the final solution, anticipating business requirements and suggesting alternative approaches when necessary.
- Created marketing strategies for a top, frequent flier loyalty program, building consumer clusters and models looking to predict response to a given promotion, designing experimental campaigns, and scoring prospects to optimize new offers.
- Solved a payment reversal risk model problem by applying outside-the-box thinking on a subprime portfolio at a mid-sized US bank. Saved the client a minimum of $2.4 million per year, and provided the breakthrough in our sales process.
- Built a solution for a top 10 US bank in the fraud space that enabled our client to determine which credit card numbers may have been compromised using a skimming device at a gas pump.
- Collaborated with software engineers in developing, prototyping, and integrating state-of-the-art statistical and machine learning algorithms into enterprise analytics software.
Technologies: Machine Learning, Consulting, Team Leadership, Information Retrieval, Cluster Analysis, Predictive Analytics, Text Analytics, Natural Language Processing (NLP), Marketing Analytics
Felice Rando
Data Scientist Developer in Seattle, WA, United States
Member since April 19, 2021
Felice is an expert in machine learning, predictive analytics, and big data, having built dozens of solutions across multiple verticals, including finance, fraud, telecommunications, and pharmaceutical research, deriving business value from raw data. He has served both as a practitioner and analytics executive in organizations such as OECD, the University of Washington, First Data, ID Analytics, and multiple start-ups. Felice has a Master's degrees in theoretical and applied statistics.
Felice is now available for hire
Portfolio
- The Spur GroupNatural Language Processing (NLP), Keras
- Data Science EvolutionBayesian Inference & Modeling, Data Mining, Jupyter Notebook, Python, SQL
- SteppeChangePython, Apache Hive, Spark SQL, Customer Segmentation, CatBoost, XGBoost...
Experience
- SQL 15 years
- Predictive Analytics 15 years
- Machine Learning 12 years
- Python 10 years
- Big Data 8 years
- XGBoost 6 years
- Natural Language Processing (NLP) 6 years
- Customer Segmentation 5 years
Availability
Part-time
Preferred Environment
Jupyter Notebook, Python, R, Slack, VS Code
The most amazing...
...tool I've built was a series of models that improved the lives of millions of telecommunications customers by creating more strategic, personalized offers.
Employment
Experience
- Predict the Likelihood of Exceeding Mobile Phone Allowances
I built a model for telecommunication marketing to predict the likelihood of an individual cell phone user to exceed their threshold at an early point of the 30-day billing cycle, scorable at any point in the cycle (including very early in the cycle, before day ten), to be able to make a new offer or an add-on offer to the customer. Significant time-aware feature engineering was used, and the models were largely built in XGboost.
- Review flagging
Built a Keras/NLP model to determine whether a product review on nutritional supplements should be flagged for various violations, including profanity and low quality reviews and/or medical advice on a challenging dataset containing multiple changes in business practices.
- Compensation Analysis
Used a Bayesian Network to model the distribution of various components of compensation (specifically salary and bonus) using limited inputs, such as geographic location, years of experience, job function and title, and education. Returned a distribution of which the 5/25/median/75/95 percentiles were of interest to the end user.
- Customer Loyalty and Segmentation
Examined and analyzed the distributions of two major loyalty programs that were to be combined during a merger to ensure data integrity, reduce overlap, and proposing customer segments. Gave feedback to the engineering teams regarding data transformations to render the distributions as equivalent as possible. Built models looking to predict probability of response to a given promotion, designing experimental campaigns, and scoring prospects in order to prioritize work and optimize new offers.
- Fraud Prevention
Built solutions to determine identity fraud during consumer applications. Created features out of interconnected applications using identity elements across a consortium of lenders and service providers including banks, utilities and telcos. Identified and evaluated third party data providers to provide additional lift to models. Used data to create secondary models such as orthogonal credit scores, and to create additional proofs of concepts such as identifying tax fraud and fraud rings.
- Attrition
Built solution identifying required data and models to determine customer subscription churn for a customer in the identity theft prevention space. Combined results with estimation of customer value to route inbound calls to appropriate level customer service representatives.
Skills
Languages
SQL, Python 3, SAS, Python, ROther
Logistic Regression, Statistics, Cluster Analysis, Machine Learning, Predictive Analytics, Data Mining, Gradient Boosted Trees, Explainable Artificial Intelligence (XAI), Artificial Intelligence (AI), Natural Language Processing (NLP), Customer Segmentation, Text Analytics, Marketing Analytics, Big Data, Churn Analysis, Data Visualization, Loyalty Management, Customer Lifetime Value, Random Forests, Customer Research, Experimental Design, Generalized Linear Model (GLM), Demography, Time Series Analysis, Multivariate Statistical Modeling, Principal Component Analysis (PCA), Probability Theory, Statistical Methods, Bayesian Inference & Modeling, Applied Mathematics, Executive Presentations, Fraud Prevention, Credit Risk, Leadership, Mentorship & Coaching, Team Building, Consulting, Team Leadership, Information Retrieval, Time Series, Feature Engineering, Loyalty Programs, Compensation, Sentiment Analysis, Reviews, Feature Analysis, Health Insurance, Pharmaceutical Marketing, Fraud Investigaton, Decision Trees, Financial Services, Churn Management, Customer Data, AWSTools
AutoML, DataRobot, Slack, Microsoft Teams, VS Code, Spark SQLLibraries/APIs
Keras, CatBoost, XGBoost, Scikit-learn, NumPy, TensorFlowParadigms
ManagementPlatforms
Jupyter NotebookStorage
Apache HiveIndustry Expertise
Telecommunications, Marketing
Education
- Postgraduate Certificate in Executive Perspective for Scientists and Engineers2012 - 2013University of California - San Diego, California, USA
- Master's Degree in Applied Statistics1999 - 2001Macquarie University - Sydney, Australia
- Master's Degree in Statistics1992 - 1998La Sapienza - Rome, Italy
