Hannah Lyon Email & Phone Number

New York, Ny, Us

Ridgefield, Ct, Us

● Spearheaded requirements necessary to build a suite of predictive models that would lead to important KPI growth through development of a data cleaning dashboard in Streamlit and Google BigQuery● Significantly contributed to the data modeling, MDM, and ETL processes necessary to build curated datasets for downstream analytics use cases including feature engineering pipelines and customer segmentation analyses● Retroactively initiated an experimentation framework for the organization’s online platform using ML and causal inference models to determine which clients should be targeted regionally to drive up adoption rates.

● Generated an analytics dashboard by leveraging Facebook Prophet and two years of eCommerce datasets to create a sample forecasting model, accurately predicting store monthly sales for the next six months for sales forecasting● Built fuzzy matching tool using (AI) Natural Language Processing (NLP) tools including TF-IDF for data cleaning which increased productivity by saving two hours of manual data labeling each month

West Chester, Pennsylvania, Us

● Owned a graph-based, collaborative filtering product recommendation model that drove two strategic initiatives in the personalization of the company’s B2C e-commerce site, decreasing expenses● Partnered with eCommerce team to improve conversion rates by ~5% by developing a (ML) gradient-boosted tree model that identified critical site features for driving sales and validated earlier findings from the testing team● Designed features from multiple data sources using Python and SQL for improving an existing airtime attribution model which improved the model by ~5-10% to better understand customer purchasing behavior.

San Francisco, Ca, Us

● Developed a data-driven marketing attribution model using 4th-order Markov chains, enabling the organization to gain valuable insights and optimize marketing spend by saving ~3% in marketing expenses● Created a customer propensity scoring model utilizing (ML) gradient boosting (xgboost) to identify critical site features, which was then leveraged by the digital team to enhance the website and improve conversion rates● Combined SQL and Tableau for ad hoc data analysis of payment methods visualized in a dashboard that delivered data-driven insights which were leveraged by the digital marketing team on a daily basis● Utilized Pytorch to train a neural network to produce product embeddings, which were leveraged for a recommendation system on website product pages - increasing page views for the recommended products by ~2-6%● Leveraged an artificial neural network in statistical modeling of repeat purchase behavior which accurately predicted second purchases with a 97% accuracy rate, enabling the organization to improve customer retention by ~4%

Stamford, Ct, Us

• Pulled data from the Twitter API to create social media sentiment reports and benchmarking tools used by the bank daily• Developed predictive model for customer attrition using logistic regression in Python packages scikit learn and statsmodels which served as a proof of concept for the bank’s integration of Python into its technology stack

• Analyzed bus networks in Brooklyn, New York to recommend changes to optimize bus routes• Visualized and explored GTFS and network data in Python to support academic research

San Jose, Ca, Us

• Entered data into company database and aided data quality assurance• Coordinated outreach to food bloggers and YouTubers to arrange partnerships

Master Of Science - Ms, Data Science

Bachelor Of Science - Bs, Data Science With A Concentration In Economics

High School Diploma