Geometric features extracted from point clouds are one of the intricate tools which can be further utilise to enhance the classification and segmentation problems. Point cloud being an unstructured data can be very problematic for performing operations such as object detection and feature extraction, using theses derived information (such as: planarity, verticality, Omnivariance, roughness etc.) can help us to better understand the quality of datasets. In this study, we have demonstrated these Eigen value based analysis using an open source platform over a photogrammetric point cloud using PCA. These feature classes could be very useful in designing more robust ML or DL algorithms using the existing information extracted from the point cloud. An experiment has been performed over a Hill-top fort point cloud to create multiple geometric features and a binary class of built up surface and natural surface component. We also provided an understanding of these features from a mathematical standpoint. Somewhat quality of a datasets can be judged from these tools for minimising the error in later processing and 3d modelling stages.