Masters of Science in Statistics
Permanent URI for this collection
Browse
Recent Submissions
- ItemForecasting The Effect of Social Media Usage on The Academic Performance of Kampala International University Students Using the Prophet Model(2026) Nabunya ZaharahThis study investigates the effect of social media usage on the academic performance of students at Kampala International University. The study focuses on key usage dimensions, including frequency, duration and purpose, alongside temporal factors such as seasonality, changepoints and holiday effects. A sample of 371 students was used, with data collected as weekly observations of academic performance measured through examination scores from January 2020 to the 2025–2026 academic period. A quantitative research approach was employed, integrating time-series forecasting and regression-based analysis. The Prophet model was applied as the primary forecasting technique, alongside ARIMA and multiple linear regression for comparison. Model performance was evaluated using MAE, RMSE and MAPE. The results indicate that social media usage has a statistically meaningful but modest negative effect on academic performance. Social media usage (β_1= −0.0077) and frequency (β_2= −0.0013) show negative associations, while digital literacy (β_3= = 0.0039) and perceived social media effects (β_4= 0.0053) exhibit small positive contributions. Temporal analysis reveals seasonal variations and structural changes corresponding to academic events such as examinations and semester transitions. Forecast results suggest that academic performance remains relatively stable over time, with predicted weekly mean scores fluctuating around the mid-70s range. Holiday effects are short-term and do not produce lasting impacts. In terms of model performance, the Prophet model outperforms ARIMA and regression models, achieving MAE = 3.185 and RMSE = 4.124, compared to ARIMA (MAE = 3.503; RMSE = 5.195) and regression (MAE = 3.923; RMSE = 4.915; MAPE = 5.67%). The findings confirm that social media usage is a significant predictor of academic performance. The study emphasizes the importance of controlled and purposeful social media use in academic settings.
- ItemPredicting the Effect of Foreign Direct Investment on Macroeconomic Indicators in Somalia(2026) Abdulahi Dahir IsmailThis study investigates the influence of Foreign Direct Investment (FDI) on Somalia s key macroeconomic indicators, specifically Gross Domestic Product (GDP), unemployment, and inflation, using Decision Trees and Generalized Additive Models (GAMs) to capture both linear and nonlinear dynamics. The research employs 32 years of secondary data (1991 2023) from the World Bank Development Indicators. Decision Tree analysis identified critical FDI thresholds influencing unemployment and GDP, with the model showing strong predictive accuracy for unemployment (R² = 0.851) and GDP (R² = 0.837), while performance was weaker for inflation (R² = 0.780). GAMs revealed a significant nonlinear effect of FDI on inflation (pseudo R² = 0.401), suggesting that changes in investment influence price levels in a non-straightforward manner, whereas the effects on GDP and unemployment were modest. Model comparison indicated that Decision Trees better capture segmented patterns in unemployment, whereas GAMs are more effective at modeling smooth nonlinear trends in inflation. Forecasts for the next ten years predict steady growth in FDI and GDP, a gradual rise in unemployment, and a decline in inflation, highlighting potential areas for policy intervention. The findings support targeted FDI policies, particularly toward labor-intensive sectors, and emphasize the importance of selecting appropriate predictive models to inform macroeconomic planning. Overall, the study demonstrates that FDI can be leveraged to improve Somalia s economic stability, provided that threshold effects and nonlinear interactions are carefully considered.