Time-Series Forecasting Model Replication in Power-Distribution Equipment Systems of Uganda: A Methodological Evaluation
Abstract
This study aims to replicate a time-series forecasting model for power-distribution equipment systems in Uganda, focusing on methodological evaluation and yield improvement. A replication study will be conducted using a time-series forecasting model (e. g. , ARIMA) with an uncertainty statement of ±5% confidence intervals to evaluate the reliability of predictions in Ugandan power systems. The analysis revealed that the model's predictions for yield improvement were consistently within ±5% of actual outcomes, indicating high predictive accuracy and robustness. The replication study confirmed the original findings but also highlighted areas where further refinement is needed to improve prediction precision in Ugandan power distribution environments. Recommendations include incorporating real-time data updates into the forecasting model and conducting sensitivity analyses for different operational conditions. The maintenance outcome was modelled as Y₈ₓ=₀+₁X₈ₓ+uᵢ+₈ₓ, with robustness checked using heteroskedasticity-consistent errors.
Key Points
Objective
The study aims to assess the effectiveness of a time-series forecasting model for power-distribution equipment in Uganda.
Methods
- Conducted a replication study using the ARIMA time-series forecasting model.
- Evaluated model predictions with a ±5% uncertainty statement.
- Modelled maintenance outcome using a specified function.
- Performed robustness checks for heteroskedasticity.
Results
- Model predictions were consistently within ±5% of actual outcomes.
- Confirmed original findings of predictive accuracy and robustness.
- Identified areas needing refinement for enhanced prediction precision.