Identifying a potentially erroneous device in an internet of things (IoT) network

Examples described relate to identifying a potentially erroneous device in an IoT network. In an example, data from a device in an IoT network. The data may be analyzed against a classification of previous data from the plurality of devices, wherein the classification classifies the previous data as one of an erroneous data, a potentially erroneous data, or a normal data. If the data from the device represents the erroneous data, the device may be included in a set of potentially erroneous devices. A cluster of the device may be determined in a cluster classification. If the device belongs to the erroneous cluster in the cluster classification, associated devices in the erroneous cluster may be added to the set of potentially erroneous devices. If a threshold amount of devices in the set of potentially erroneous devices is higher than a pre-defined value, a notification may be generated.