Reserach Project Using Machine Learning

My proposed work includes three phases Phase I : Balancing the given stream dataset using resampling technique Phase II : Constructing indexing structure of ensemble model 1. Phase III: Updating model with new arrival of data stream Phase I 1. To achieve first objective map reduce based ensemble model will be constructed with base classifier as VFDT Model using resampling technique. 2. Model should handle imbalance and concept drift issue simultaneously. 3. Imbalance issue : This model will give the output as balanced dataset and it should able to handle all data difficulty factors like small disjuncts, overlapping classes and boundary classes using Distributed HDBSCAN algorithm. 4. Should handle multiclass classification using ensemble model. 5. Model should be evaluated on following synthetic dataset and real world datasets. Synthetic Datasets : SEA(3-class),Hyperplane (2-class),RBF (4-class): can be generated using MOA tool. Real World dataset : CoverType(multiclass – 7 class , Available on UCI machine learning repository),Malicious URL dataset (multiclass available on Kaggle .github) 6. Comparing model with existing techniques GRE,AWE,SMOTE etc.