Model comparison on a grouped test set of 457 advertisements.
Python · pandas · scikit-learn
Tech Job Salary Prediction
Explored job advertisements, engineered features and compared Linear Regression, Random Forest and tuned Gradient Boosting. Matching advertisements were kept in the same split to reduce leakage.
Methods, findings & limitations
Assessed MAE, RMSE, R² and country/job-level residuals. Linear Regression had the lowest test RMSE; model rankings differed from training cross-validation. Unverified currency and pay-period units limit practical salary recommendations.
Completed notebook · GitHub link to be added when published.
Global vaccination trend chart recreated from the project dataset.
Tableau · Data cleaning · Regression interpretation
Global COVID-19 Vaccination
Built an interactive dashboard covering global vaccination progress, daily activity, income-group differences, country coverage and GDP per capita.
Methods, findings & limitations
Reported global full vaccination coverage of 64.93 per 100 people as of 14 August 2024. GDP analysis covered 189 countries (R² = 0.328). Historical data and differing reporting dates require care; association does not establish causation.
Validation summary: 1,088 patient, lesion and joined records.
PostgreSQL · Joins · Data validation
Skin Cancer Data Analysis
Validated joined patient and lesion records, examined identifiers and missing measurements, and explored diagnosis patterns, demographics and lesion characteristics.
Methods, findings & limitations
Produced a SQL script and findings report, separating descriptive patterns from causal or diagnostic claims. Documented missing measurements and repeated lesion identifiers.
