This project entailed finding the type of fault that existed in roller bearings. The dataset contained values of acceleration in X and Y direction. At first, we denoised the dataset using Wavelet based soft thresholding and Savitzky Golay Filter. Then, we extracted essential features from the data by dividing it into different classes. After that, we selected the essential features using Principal Components Analysis. Finally, we identified the type of fault in roller bearings using k-means and Dendrogram Clustering with an accuracy of 80%. We used Python programming language for writing code.