Data Scientist

Languages/Tools: SAS, SQL, Python, R, Tableau

Analytics: Supervised Machine Learning Algorithms, Unsupervised Machine Learning Algorithms, Data Visualization, Exploratory Data Analysis, Scikit-Learn, Numpy, Pandas

Contact

• Email: trevormaxwell08@gmail.com
• Phone: (612)-462-6372
• Location: Philadelphia, PA 19103

Education

• Master of Science, Data Science
  • Bellevue University (2022 - 2024)
• Nanodegree Certification, SQL
  • Udacity (2020)
• Nanodegree Certification, Data Analyst
  • Udacity (2020)
• Bachelor of Arts, Physics
  • Gustavus Adolphus College (2013 – 2017)

Professional Background

Senior Analyst @ Aetna, A CVS Health Company (Jan. 2022 - Present)

• Develop automated reporting solutions using SAS to improve decision-making processes for stakeholders
• Design and implement SQL tables to deliver insights and solutions for business processes
• Manage and transmit data to data scientists for ingestion by their predictive models leading to reduced member complaints
• Work independently on reporting solutions using a development, test, and production methodology

Data and Analytics Consultant @ Aetna, A CVS Health Company (Nov. 2020 - Jan. 2022)

• Transformed internal customer needs into technical requirements to build reporting solutions
• Collaborated with cross-functional teams to develop and transfer Medicare Advantage eligibility reports via sFTP
• Directed the submission of Plan Communication Transactions to CMS and managed the associated Submission Helpdesk inbox
• Designed recurring and ad hoc reports using SAS and SQL

Analyst, Quality Engineering @ ReSound (Apr. 2020 - Sep. 2020)

• Collected, processed, and organized medical device data for quality engineers and manufacturing engineers
• Created weekly, monthly, and ad hoc analyses using QlikSense
• Presented analyses to global management and engineering teams
• Developed key performance indicators to monitor device quality

Coordinator, Quality Engineering @ ReSound (Aug. 2018 - Apr. 2020)

• Managed a project involving computer software validation, decision-making processes and quality-related data
• Applied root cause analysis to investigate and trend defected devices
• Selected Continuous Improvement captain leading a team of assemblers to seek improvements on their daily tasks by applying Lean manufacturing principles

Data Science Projects

• Click the header to view the Github Repository.

End-to-End Data Pipeline: ETL from API to SQL with Python for Tableau Visualizations

Developed an end-to-end ETL solution using Python to extract Copa America data from four different Sportmonks API endpoints. This data was transformed and combined into clean datasets and loaded to two different SQL tables for efficient querying and reporting. The final output was visualized in an interactive Tableau dashboard, allowing for insightful analysis of key metrics and match results.

Time Series Forecasting for National Sales

Developed time series forcasting models to sales data for a national company using Python. ARIMA, SARIMA and XGBoost models were trained to determine which model was most accurate. Based on the RMSE, the XGBoost model fit best. Forecasting sales can aid organizational operations planning as well as the development of key performance indicators.

Customer Segmentation Using K-Means Clustering

Used Python to develop unsupervised machine learning model that created customer segments. The model trained and deployed was K-Means Clustering. The segments can be used to build marketing strategies that generate more customers as well as improve current policies or generate new policies.

Marketing Offer Acceptance Prediction

Supervised machine learning models were developed using Python to predict which customers are most likely to respond to an ad campaign. The model selection phase determined the Random Forest model had the best outcome based on the accuracy and precision scores. Predicting who is most likely to respond to an ad campaign allows for the business to have better targeted marketing strategies.

Collecting Real-Time Weather Data Via OpenWeatherMap API

Used PyCharm to develop a connection to the Open Weather Map API and output current weather based on the inputted city and state or zipcode. Connecting to the API provides the current weather for the input location in real-time.

Heart Failure and Mortality Prediction

Used Python to predict mortality with supervised machine learning models. Based on accuracy and recall scores, Random Forest Classifier performed best. Hospitals and providers could use this model to provide additional care and reduce follow-up time for patients at high risk.

Data Gathering from SportRadar API

Connected to the SportRadar API using Python to gather depth chart data for running backs in the NFL from 2006 - 2017. Gathering data from an API crucial for data science and software development. APIs are cost-effectiveness and vital for accessing real-time information and scalability.

Follow-Up Treatment Prediction: Inpatient or Outpatient Visit

Developed supervised machine learning models in Python to predict if a patient’s next treatment should be an inpatient or outpatient visit. Four models were trained to the data, and the Random Forest model performed best based on the accuracy score. This prediction could aid the hospital in planning/staffing for the visits as well reducing the decision-fatigue on providers.

Hospital Readmission Prediction

An exploratory data analysis and logistic regression model were performed using Python on patient data which included the time spent in hospital of the initial visit, diagnoses, number of medications administered, number of procedures and lab procedures perfomred, and lab results among others. The analysis concluded that with each additional medication administered increased the odds of the patient being readmitted by 1.01%.

Reducing Preventable Asthma-Related Hospital Visits

Used R to collect, clean, and combine data from three separate sources: preventable asthma hospital visits, emergency department hospital visits for asthma, and the average annual temperature in the county in which the visits occured. Linear and Multiple Linear Regression models were fit to the data, and the Multiple Linear Regression Model fit best based on the adjusted R-squared value. The analysis concluded that as the average annual temperature increases, the rate of preventable asthma visits slightly increases.