In this age of big data, system produce massive amounts of data, that further derives more data, and even propagated to different clouds, forming a data propagation network, that is difficult to analyze, known as Data‟s lineage. Data‟s lineage is the derived data, which sometimes users wants to delete. There may exist several reasons that users want systems to delete certain data with its lineage. Removing noise and incorrect entries from lineage allows the recommender system to give only useful recommendations. Also, one may want the recommender systems to forget their data, to avoid privacy risks. This paper focuses on transforming the machine learning algorithms to forget the data (or say its lineage) and also Machine learning level user authentication to provide users better security mechanisms and ensure learning algorithms to maintain privacy and hassle free data. We present a comprehensive approach of transforming learning algorithms to a special form i.e. called Statistic Query Learning. This approach uses small number of aggregation, i.e. faster than retraining recommender system from scratch. And also this approach can be applied to all levels of machine learning algorithms. © 2016, International Journal of Pharmacy and Technology. All rights reserved.