Applications that run on mobile phones shaped a trendy lifestyle for many users nowadays. This led to a significant growth in the proportion of data traffic, relative to voice traffic, to be delivered in the mobile phone network such as GSM. Traditionally the underlying radio resources in GSM networks for data and voice traffic were allocated by some predefined traffic policy which was manually configured. The allocation may not be most accurate for the fact that demands for data traffic fluctuate largely and temporally. A new resource planning scheme is desired that can dynamically adjusts the resource allocations according to the latest information of the traffic statues. In order to facilitate such dynamic resource allocation, a resource management system is proposed in this paper. Data mining is used to derive rules and extract traffic patterns that reveal critical information for setting values in resource planning. Empirical testing data are used in experiments that demonstrate the efficacy of the data mining techniques.