Abstract\n Objective: The main objective of this study is to develop a robust, scalable, efficient classification algorithm for imbalanced data which is based on the classification of enzymes and lower approximation of subset and can handle large amount of data.It can also be used for preprocessing of imbalanced data.\n Methods and Analysis: Extrinsic or Data level and Intrinsic or algorithmic level solutions exists to tackle the problem of imbalanced training sets. The methodology adopted in this study is Data level solutions, it is those that try to resample and balance the dataset by increasing the artificial-instances in the smaller class by constructing new artificial samples, known as over sampling.This study proposes a new algorithm through construction of new samples using the synthetic minority oversampling technique hybridized with the application of a new algorithm called Enzyme-Computation which is an editing technique based on enzyme-classification and lower approximation of a subset. We show that any data can be classified by Enzyme-Computation algorithm similar to as enzymes are classified into 6 classes.\nFindings: The proposed algorithm called Enzyme-Computation has been experimentally evaluated with other preprocessing algorithms, validated and supported by comparative validation and shows good results. For comparative analysis we have taken 22 datasets from the UCI repository. The other algorithms chosen for comparison are: SMOTE, SMOTE-ENN, SMOTE-TomekLinks, Borderline-SMOTE1, Borderline-SMOTE2. We applied the hypothesis testing technique that provides support to the analysis of the results.We use different types of tests.For multiple comparisons; we use the Iman-Davenport test to detect statistical differences among a group of results.To show how good a method is with respect to its partners we consider the average ranking of the algorithms. The ranking of the Enzyme-Computation algorithms on each data-set selected for this study is also studied.In order to compare the results; we used a multiple comparison test to find the best preprocessing algorithm. We have observed that the best ranking is obtained by our proposal-Enzyme-Computation and the two last positions correspond to Borderline-Smote1 and Borderline-Smote2.\nApplication/Improvements:The novelty of this proposal is that the quality of the new synthetic instances is evaluated using Enzyme-Computation Algorithm .This evaluation allows us to include only those artificial instances that are within the lower approximation of the minority class.Further this algorithm can be enhanced based on new concepts developed for physico-chemical properties of Enzymes and its classification.