Experiments and simulations in materials science and engineering often generate prodigious quantities of data. Extracting information from this data turns out to be more challenging than may at first appear, prompting efforts to create innovative ways of analyzing “big data.” I will provide an overview of my own adventure in data-driven materials research and focus in on a few methods and example applications that have proven to be especially productive. The first deals with the inference of failure criteria for individual microstructure features from databases of individual failure events. The second concerns mining and analysis of image data from the open literature to gain new insight into materials without performing any new experiments or simulations. I will conclude with some thoughts about the future of data-based methods in materials science and engineering.