Milling is one of the most crucial processes in machining. Every industry demands a stable milling process for a smoother finish and material cost reduction. Chatter is a vibrating phenomenon which affects the workpiece's quality, its dimensional accuracy, and tool life. It is required to classify the chatter phenomenon to devise an effective chatter prevention strategy.

Several classification strategies are being used, including frequency and time-related strategies. Since the chattering phenomenon is a frequency-based phenomenon so a frequency-based feature set can be of vital importance. However, frequency-based strategies have a problem of noise. The noise problem can be addressed by combining frequency and time-domain methods.

Thus, a hybrid approach based on the frequency and time-based feature set is developed and used in conjunction with k-means-based unsupervised learning to come up with a practical but reliable classifier. The proposed classifier algorithm offers good performance, clearly distinguishing between chatter and stable conditions.

Based on the chatter classification in this work, it is possible to identify thresholds for chattering detection. It is essential to mention that the thresholds obtained from this work will only be useful for the machine and tool used in the experiments and will not be of use for other machines and need more investigation.