The early detection of a stuck pipe during drilling operations is challenging and crucial. Some of the studies on stuck detection have adopted supervised machine learning approaches with ordinal support vector machines or neural networks using datasets for “stuck” and “normal”. However, for early detection before stuck occurs, the application of ordinal supervised machine learning has several concerns, such as limited stuck data, lack of an exact “stuck sign” before it occurs, and the various mechanisms involved in pipe sticking. This study acquires surface drilling data from various wells belonging to several agencies, examines the effectiveness of multiple learning models, and discusses the possibility of the early detection of pipe sticking before it occurs. Unsupervised machine learning using data on the normal activities is a possible advanced method for early stuck detection, which is adopted in this study. In addition, as a countermeasure to another concern that even normal activities involve various operations, we apply unsupervised learning with multiple learning models.