We consider the problem of designing a tour for a mobile element in data producing sensor networks. The tour is designed to visit a subset of the nodes, chosen for their centrality in the network. This way the sensors that are not visited by the mobile element will need to transmit their data wirelessly. This may require several hops and therefore may reduce the lifetime of the network. The most common optimization objective for these data gathering problems is to minimize the amount of wireless transmission. For networks with relatively uniform density of nodes, there are several heuristics that work well in practice. However, if there are nodes that are placed far from the central locations of the network, then most proposed algorithms will end up designing a tour that may be skewed towards the outlier nodes. In this work we quantify the effect of outliers in the design of data gathering tours in wireless networks, and propose the use of an algorithm from data mining to address this problem. We provide experimental evidence that the tour planning algorithms that takes into account outliers can significantly improve the solution.