ATLAS is a general-purpose detector at the Large Hadron Collider (LHC), and it is equipped with two sets of special detectors, called Roman Pots, ALFA and AFP. The aim of this research work is to analyse the different classes of cluster hits and hence deepen the knowledge on how protons form the cluster hits in the AFP detector. In the years 2016-2018, ATLAS Collaboration took data of diffractive events with the newly installed Forward Proton detectors and since these devices are relatively new, there is a lot about their performance to be understood. The data used here is from 2017 low pile-up runs. To analyse these data, C++ and ROOT data analysis framework were the used tools. Methodology was as follows: first the basic properties of hits formed in the AFP detector were studied, then after making some event displays, I categorised various cluster shapes. Comparing some cluster shapes, it was observed that some shapes are more probable to occur than others. Moreover, it was concluded that cluster reconstruction in AFP should correctly recognise clusters in about 99% of the cases. This would help identify the origin of the cluster hits and lead to indications on how algorithms should form clusters.