A Dynamic Integrated Classification Algorithm Based on Big Data Environment

With the developing of big data application, classification algorithm has been expanded to distributed datasets from the single dataset. So a dynamic integrated classification algorithm based on big data environment was proposed. This algorithm gain integrated classifiers of high classification accuracy for each local dataset, and dynamically generate the recognition model according to the distribution characteristics of local samples to be tested. In the application process, after numerous new sample data join the datasets, the classifier performance will drop gradually. By aiming at the above problem, this algorithm will retrain the classification model in the dynamic expansion process of datasets. According to the experimental results, the algorithm proposed in this paper has high classifier training performance and classification accuracy. At the same time, it also possesses high adaptive capacity when faced with dynamically changing distributed datasets.