The Propelled analytics of data streams is rapidly turning into an important territory of processing the complicated data as the quantity of requesting such applications grows continuously. Web based mining when the input data advancing over period is getting to be most important center problem. In todays real world the data is not stationary but non stationary. The properties of attributed are changing over time. As a result the data stream has concept drift and noise which affects the performance. The aim of the paper is to first present an overview of the challenges in the data streams, followed by the measures to improve the factors that affect the performance. © BEIESP.