The Internet of Things (IoT) revolutionized maintenance technology by allowing machines to communicate among themselves over the Internet in real-time. The very focus of Industry 4.0 towards automation and data exchange calls for the application of IoT in industry and for control, supervision, and maintenance of industrial equipment based on sensor data. The goal of this paper is to test several unsupervised machine learning algorithms for detecting anomalies in unlabeled (raw) data collected from a real e-coating plant that uses IoT technology. Unsupervised learning methods are advantageous because they do not require labeled data and benefit early stage anomaly detection in industry in general. The dataset used for training and testing is composed of real-time data gathered from the plant’s sensors over a period of 15 days. For the purpose of anomaly detection, we used the following unsupervised machine learning algorithms: Interquartile Range (IQR), Isolation Forest, and Elliptic Envelope. The algorithms were compared with each other by the time required for training and the total number of detected anomalies. According to the observed results, it was shown that IQR needs much less time for training and detects three times more anomalies compared to the other two algorithms.