Semi-supervised learning plays an important role by combining both labeled and unlabeled data, bridging the gap between supervised and unsupervised learning. In today’s fast-changing data streams, an efficient classifier is essential to adapt to evolving patterns. It is also important to address challenges like concept drift and concept evolution to maintain consistent model performance. However, handling unlabeled data in semi-supervised data streams can be challenging, as it requires significant resources and can be costly. This paper introduces an Incremental Workflow Model (IWM) that effectively handles the underlying issues of the data stream in a semi-supervised environment while using minimal labeled information. The proposed model is designed to be highly adaptable, ensuring it can handle the concept drift and concept evolution challenges. One-Class Classifier (OC_Classifier) with a proposed confidence score is used to identify the novel data and to predict known data in the stream. Further, MicroClusters are employed to reduce the memory overhead in the Ensemble Model. Experiments are performed on benchmark datasets, and results are analyzed to demonstrate the effectiveness of the proposed workflow model.