Data streaming is the transmission of a continuous data stream which is often fed into stream processing software to produce insightful data. A collection of data elements arranged chronologically make up a data stream. The two methods used to classify data streams are single and ensemble classification. The single classification technique is quick and uses less memory for processing, but as the number of unknown patterns or samples rises, its efficiency declines. The ensemble technique can be utilized for two main reasons. Compared to a single model, an ensemble model can perform better and result with accurate predictions. Ensemble learning (EL) generates various base classifiers form which a new classifier is produced that efficiently performs than other traditional classifiers. In addition to the algorithm, hyperparameters, representation and training set, these base classifiers may differ in the type of classification. A dynamic ensemble learning (DEL) is a sort of EL algorithm which automatically selects the subset of the ensemble members while making the prediction. The primary benefit of DEL is improved predictive accuracy when compared to normal EL. This paper study and analyse various dynamic ensemble classifier model for data stream classification. The identification of merit and demerit of these methods help to understand the available problems and motivate researchers to find out new solutions for the listed problem.