This document discusses data mining techniques for analyzing functional magnetic resonance imaging (fMRI) data. It begins with an introduction to neuroscience and fMRI. Common approaches to analyzing fMRI data include single-voxel analysis and region of interest (ROI) analysis, each with pros and cons. The document then describes a cluster-based analysis (CBA) method that forms clusters of correlated voxels to increase signal-to-noise ratio. CBA allows discovering activations across the whole brain while controlling for multiple comparisons using false discovery rate procedures developed for fMRI data. In summary, CBA is presented as a hybrid approach that combines advantages of single-voxel and ROI analyses to better analyze fMRI data.