This dataset provides experimental performance measurements for O-RAN scheduling xApps evaluated in an end-to-end O-RAN–compliant testbed. It accompanies the paper “xApp-Driven Throughput and Demand-Aware Resource Allocation in O-RAN Systems” (submitted to ICC 2026). The dataset compares a custom Throughput and QoS-aware radio resource management (RRM) xApp, MUMT (Minimize Unmet - Maximize Throughput), against two benchmark scheduling algorithms, Round Robin and Proportional Fair. Each xApp is deployed independently and evaluated under identical experimental conditions, including different traffic demand levels and pathloss-based scenarios. Experiments were conducted on an O-RAN-compliant end-to-end (E2E) B5G testbed using an open-source gNB, UEs, and near-RT RIC. Measurements were collected at the UE side and include downlink and uplink performance indicators, radio link metrics, and system-level statistics, aggregated over 1-second intervals. The dataset is organized by traffic demand, pathloss scenario, and experimental runs, and is provided in raw CSV format. It is intended to support research on O-RAN xApp design, RAN intelligence, scheduling policy evaluation, and performance benchmarking.For further details, please refer to the README.md file located in the compressed file. If you use this dataset in your work, please cite the associated publication.