Bezawit Assefa
Data Scientist specializing in machine learning, time series
Profile in progress
Bezawit is building their profile!
The Ethiopia Financial Inclusion Project analyzes and forecasts financial inclusion trends across Ethiopia. Using historical data, event impact modeling, and scenario-based forecasting, the project explores factors affecting account ownership and digital payment usage. An interactive Streamlit dashboard allows stakeholders to visualize trends, assess the impact of key events, and explore future projections for 2025–2027.
Key focus areas: data enrichment, EDA, event modeling, forecasting, and interactive visualization, highlighting opportunities and risks for financial inclusion in Ethiopia.
0
26
This project analyzes Google Play Store reviews from three major Ethiopian banks—Commercial Bank of Ethiopia, Dashen Bank, and Bank of Abyssinia—to uncover customer sentiment and common user experience issues in their mobile banking apps.
Using Python, I built a data pipeline to collect, clean, and analyze user reviews, applying sentiment analysis and theme extraction to identify key topics such as transaction issues, UI usability, and customer support feedback. The insights are presented through an interactive dashboard built with Streamlit, allowing stakeholders to explore ratings, sentiment trends, and review themes in real time.
This analysis helps product and banking teams better understand customer feedback, prioritize improvements, and enhance the overall mobile banking experience.
0
29
This project analyzes historical car insurance data to identify low-risk customer segments and build predictive models for claim severity and premium optimization. Using Python, I performed exploratory data analysis, hypothesis testing, and machine learning modeling to uncover patterns in customer behavior, vehicle characteristics, and insurance claims.
Multiple predictive models, including Random Forest and XGBoost, were developed and evaluated to improve risk prediction accuracy. I also implemented data version control using DVC to ensure reproducible experiments and proper dataset management.
The analysis provides insights that can help insurance companies refine pricing strategies, reduce risk exposure, and design targeted marketing strategies for low-risk customers.
0
31
This project analyzes Brent crude oil prices (1987–2022) to detect major structural changes using Bayesian change point analysis. Using Python and PyMC, I built a probabilistic model to identify when significant shifts occurred and measure their impact. The analysis revealed a major market shift around February 2005, when oil prices moved into a higher long-term regime. These insights help investors, policymakers, and energy companies better understand market volatility and make more informed decisions.
0
43