Using Artificial Intelligence and Deep Learning to Analyze the Impact of Moroccan Investments in Africa on Trade Flows(2007– 2023)

This study investigates the impact of Moroccan investments on trade flows with 27 African countries from 2007 to 2023. After data cleaning, 295 observations were analyzed. Weak linear correlations were observed between trade flows and explanatory variables, except for Morocco’s GDP over time and African partner GDP with imports. Four Multilayer Perceptron (MLP) architectures were applied to predict exports and imports, evaluated using MSE, MAE, and R ² . The Deeper MLP achieved the best performance, with African partner GDP (PIBA) identified as the most influential predictor. Exports additionally depended on investment and political factors such as BITs and OFDI. Residual analysis revealed heteroscedas- ticity for exports and bias for imports. Overall, domestic investment drives trade flows, deeper neural networks improve predictions for complex targets, and simpler models provide better generalization, offering practical insights for policymakers and investors.