Abstract: Data mining is the science of discovering hidden patterns from data. Over the past years, a plethora of data mining algorithms has been developed to carry out various data mining tasks such as classification, clustering, association mining and regression. All the methods are ad-hoc in nature, and there exists no unifying framework which unites all the data mining tasks. This study proposes such a framework which describes a data modelling technique to model data in a manner that can be used to accomplish all kinds of data mining tasks. This study proposed a novel algorithm known as Frequent Model (FM)-Growth, based on Frequent pattern (FP)-Growth algorithm. The algorithm is used to find frequent patterns or models from data. These models will then be used to carry out various data mining tasks such as classification, clustering. The advantage of these frequent models is that they can be used as it is with any data mining task irrespective of the nature of the task. The algorithm is carried out in two stages. In the first stage, we grow the FM-tree from the data and in the second stage, we extract the frequent models from the FM-tree. The accuracy of the proposed algorithm is high. However, the algorithm is computationally expensive when searching for frequent models in high volume and high dimensional data. The reason of expensiveness is that it needs to travel all the nodes of a tree. The study suggests measures to be taken to improve the efficiency of the overall process using dictionary data structure. Keywords: Data Mining, Frequent Pattern Recognition Unified Framework, Classification, Clustering, FPGrowth tree.