Technology continues to be applied in the field of geology in various branches of science, one of which is the use of machine learning methods which are included in artificial intelligence technology. Machine learning methods able to identifying rock minerals. Rock mineral clustering is carried out to identify the distribution of the optimal number of mineral groups based on geological information held in rock drilling results data during the geological exploration stage in the Manjimup region, Western Australia. Identification of rock minerals through clustering is carried out using unsupervised machine learning with the K-Means clustering method. The data used in this research are data from the measurement of the electromagnetic spectrum in the form of Thermal Infrared (TIR) spectral data derived from rock drilling results. The spectral data used consisted of 341 parameters so that the input dimension was reduced to reduce computational complexity using Principal Component Analysis (PCA) into two-dimensional data so able to visualized more easily. Based on the evaluation results, the optimal number of rock mineral groups through the results of clustering using K-Means based on geological information is 3 groups of rock minerals