Data Streaming plays a major role due to data generated from an infinite number of sources. Stream data is the continuous flow of data from heterogeneous sources. Stream data allows the system to process the data in real-time that respond much faster than other data processing methods. Unlike traditional batch processing, stream data processing utilizes less time to process the data and provide immediate response in real-time. Deep learning models produce superior and accurate results in handling huge data streams generated from various sources with better prediction. Streaming data requires analysis of immediate insights extraction and fast decision making, which is highly possible with deep learning models. In stream classification and data stream mining, an abrupt change in the distribution of data sets over a period of time is said to be concept drift. Deep learning models detect the concept of drift and track the changes in data stream classification applications. Commonly used deep learning algorithms are the convolutional neural network, long short-term memory, general adversarial network, and active learning. Real-world application areas of stream data classification are manufacturing, healthcare, security, retail, finance, social media, internet of things, among others. Advancements in streaming data using deep learning are health monitoring systems by sensing medical stream data, Intelligent transportation systems, self-driving cars, personalized activity recognition using stream data, smart cities with privacy, Multiple topic detection, and multi-level sentiment classes from steaming sentences, and more.