Ubiquitous sensor networks gain tremendous popularity nowadays with practical applications such as detection of natural disasters. These applications collect real-time data about the atmospheric measurements from sensors that are installed in the field. In this paper we argue that traditional data mining methods run short of accurately analysing the activity patterns from the sensor data stream. We evaluate the successor of these algorithms which is known as data stream mining by using an example of an indoor ubiquitous sensor network. They measure various atmospheric values that are supposedly prone to the influences of different human activities. Superior result is shown in the experiment that runs on this empirical data stream. The contribution of this paper is on a comparative study between using traditional and data stream mining algorithms, in a scenario where different atmospheric patterns are to be recognised from streaming sensor data.