This article summarizes the methods, observations, challenges and implications for exploratory analysis drawn from two learning analytics research projects. The cases include an analysis of a games-based virtual performance assessment and an analysis of data from 52,000 students over a 5-year period at a large Australian university. The complex datasets were analyzed and iteratively modeled with a variety of computationally intensive methods to provide the most effective outcomes for learning assessment, performance management and learner tracking. The article presents the research contexts, the tools and methods used in the exploratory phases of analysis, the major findings and the implications for learning analytics research methods.