Atom Probe Tomography (APT) resolves atoms in real space and detects their chemical identity. In the realm of microscopic techniques, it has its unique place with sub nanometer spatial resolution and high chemical sensitivity. In this thesis, data driven techniques have been developed to identify nano scale chemical features from reconstructed atomic data obtained from APT experiments. Main drawbacks in present day methods, to detect nano scale features, are use of input parameters involving heuristics on part of user. Techniques developed in this work use the APT data to determine the input parameters, thereby making the process more quantitative. In particular, techniques have been developed to select optimal voxel size to calculate the concentration profile, and to select the concentration threshold using ideas from computational topology. Voxel size is selected using an error minimization technique. A framework to quantify and visualize spatial uncertainty in isosurface has also been developed. Approaches developed in this work are generic in nature and can be applied on any APT data. In this work, results have been shown for Ni based superalloy data.