In today's digital era, the online social network has become an integral part of people's social life. Online services collect a huge amount of data related to users while observing their online activities. The enormous use of data mining and sharing of the collected data has been playing a primary role in innovation and also in improving the quality of various services that have arisen in an abrupt increase of users' privacy concerns. This thesis aims to help users to identify the data that is critical regarding their privacy while they are sharing their personal information over a public platform. This work also provides an insight into the social media organization about the different age group of users who are at stake of the highest privacy Risk. For this research, the dataset Has been collected from a popular social networking site. The existing privacy preserving data mining algorithms were analysed and compared to propose an extended approach pertaining to privacy for online social network users. Analysis of existing algorithms shows the direction for this research work by focusing on the sensitive data of online social networks' users. This thesis introduces two basic approaches, namely, a Single -Classifier based approach and Multiple-Classifier based approaches based on state-of-the-art techniques for users' privacy prediction.