The effectiveness of Internet-of-Things (IoT) systems is highly dependent upon the data it collects at the front-end sensor network. If the collected data is corrupted, the performance of the back-end decision support can degrade. This research outlines a novel method for estimating the trustworthiness of data in an IoT network. The proposed ensemble-based method estimates a trust score for each data point generated by a sensor using models that assess both its consistency with other sensors within the network and its history of producing data consistent with the widespread network. One key advantage of the proposed ensemble method is its robustness to corruption from multiple sensors within the network. Numerical experiments are performed on a smart-home data set and demonstrate the ability for the proposed method to 1) identify corrupted data and 2) produce estimates of the data that are robust to simultaneous corruption in multiple sensors.