Machine Learning Business Intelligence

ESTHER NTIWAAH AWUSAH 309-***-**** ************@*****.***

LinkedIn / GitHub

Boise, ID 83703

EDUCATION

Illinois State University Normal, IL

M.S. Applied Statistics 08/2022 – 05/2024

GPA: 3.75/ 4.0

Kwame Nkrumah University of Science and Technology Kumasi, Ghana B.S.c in Actuarial Science 09/2017 – 09/2021

GPA: 3.83/ 4.0

TECHNICAL SKILLS

Programming: Python (NumPy, Pandas, Scikit-learn, Matplotlib), R Programming, SAS, PostgreSQL, SQL, MySQL Data Science: TensorFlow, Keras, Algorithms, Machine Learning, Deep Learning Business Analytics: Business Intelligence, Databases, Statistical Analysis, Data Visualization Other Tools: Tableau, Power BI, Excel, Word, PowerPoint, Git, Cloud Computing RELEVANT WORK EXPERIENCE

Pacific Source Health Plans Boise, ID

Actuarial Intern 06/2024 – 09/2024

• Analyze health-related data which improved Data Efficiency and reliability by developing a trend model for various lines of business.

• Strengthened Forecasting and Pricing Strategies by cleaning and aggregating large datasets in SAS at multiple levels, including state, LOB, and departmental levels, and developed tables that streamlined the importation and analysis of data in Excel.

• Developed and implemented Excel VBA scripts to automate the importation and updating of SAS data, ensuring consistent and accurate data availability for trend analysis and decision-making processes.

• Developed organizational goals to develop robust processes and controls to ensure data integrity, collaborating with stakeholders to troubleshoot issues, perform root cause analysis, and contribute to the validation of new SAS code for improved operational efficiency.

Voya Financial Minneapolis, MN

Actuarial Intern 05/2023 – 08/2023

• Conducted risk analysis by utilizing mortality tables, and census data. This provided key insights to underwriters and actuaries to make informed decisions about pricing, and coverage limits, mitigated risk, and ensured that the products were designed to balance competitiveness with profitability.

• Succeeded in creating customizable product plan design using Excel VBA. This allowed users to customize product plan designs, enabling them to evaluate different scenarios and assess the associated risks.

• Analyzed voluntary product data in a persistence study, identifying key retention drivers that directly informed strategic decision- making.

• Estimated baseline rate changes for multiple plan designs, providing essential insights to the pricing team. Illinois State University Normal, IL

Graduate Teaching Assistant 08/2022 – 05/2024

• Led class discussions and one-on-one tutoring sessions by conveying complex information to students to improve their understanding of mathematical concepts.

• Assisted undergraduates with research projects and term papers, meticulously guiding them through data analysis using R.

• Conducted data analysis sessions and organized tutorials, where I effectively taught complex concepts by actively listening to students' needs and adapting my teaching approach accordingly. RELEVANT RESEARCH EXPERIENCE

Time Series Analysis on Closing Price of Microsoft Stock Performed a time series analysis on the closing prices of Microsoft data from 2006 to 2017. Explored trends, seasonality, and patterns through data visualization. Different ARIMA models were developed, always looking out of parsimonious model, ARIMA (1,1,1) is the best model based on AIC and BIC, and thoroughly analyzed residuals whether my optimal model followed a white noise. Since the variance of the data was not constant, I recommended conditional variance models will be best to fit the data. Predictive Analysis on Australian Rainfall Using Deep Learning I evaluated various deep learning models, including MLP, RNN, LSTM, and GRU. I selected LSTM as the best model and developed a two- layer LSTM network with 64 input neurons. The first layer returns sequences, and the second layer produces the final output for rainfall predictions. This architecture effectively captures long-term dependencies in sequential data and extracts relevant features for classification. I used a sigmoid activation function in the output layer to interpret class probabilities and applied the binary cross-entropy loss function for measuring prediction accuracy. The Adam optimizer was utilized to dynamically adjust the learning rate and enhance convergence speed.

Contact this candidate