FEDERATED CONSUMER AND OPERATIONAL ANALYTICS FOR DEMAND-RESPONSIVE PLANNING IN DIGITAL SERVICE ECOSYSTEMS

Digital service ecosystems generate consumer-demand and operational information across distributed business units, creating challenges for coordinated planning, data privacy, and heterogeneous demand forecasting. This study proposes a federated consumer and operational analytics framework for demand-responsive planning that connects consumer-demand signals with operational capacity while retaining local data ownership. The framework integrates federated learning, client clustering, demand forecasting, Digital Twin coordination, demand responsive capacity planning, and privacy utility evaluation. A controlled synthetic dataset representing eight heterogeneous business units and 240 time periods is used for computational evaluation. The results compare local learning, centralized learning, FedAvg, and clustered federated learning. The proposed clustered federated approach achieved the lowest forecasting MAE of 9.4799, compared with 9.5721 for local learning. Demand responsive capacity planning reduced capacity shortage from 1.8225% to 0.6906% and capacity imbalance from 0.0990 to 0.0549. Privacy evaluation further demonstrated increasing forecasting error under tighter privacy settings. The findings provide computational support for connecting distributed demand forecasting with responsive capacity planning while retaining local data control.