THE ACCURACY OF FORECASTING MODELS IN FINANCIAL REPORTS OF RWANDAN COMMERCIAL BANKS

This study investigates the accuracy of forecasting models in the financial reports of Rwandan commercial banks, aiming to assess their reliability, identify factors contributing to discrepancies, and recommend enhancements. A mixed-methods approach was employed, analyzing quantitative data from financial reports (2005–2012) using statistical measures such as Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE), alongside qualitative insights from interviews with 15 senior financial analysts. Findings reveal high forecasting accuracy, with revenue and expense projections achieving average accuracies of 97.8% and 97.3%, respectively. Discrepancies were linked to inflation and exchange rate volatility, with multivariate regression confirming these factors as statistically significant predictors (p < 0.05). Advanced models, such as neural networks, outperformed traditional methods by improving accuracy by 5%. The study concludes that while current models are robust, integrating machine learning and enhancing data governance can further optimize precision. Recommendations include adopting advanced tools, capacity building, and regulatory alignment to sustain and improve forecasting reliability.