In this paper, we present "BIKED," a dataset comprised of 4500 individually\ndesigned bicycle models sourced from hundreds of designers. We expect BIKED to\nenable a variety of data-driven design applications for bicycles and support\nthe development of data-driven design methods. The dataset is comprised of a\nvariety of design information including assembly images, component images,\nnumerical design parameters, and class labels. In this paper, we first discuss\nthe processing of the dataset, then highlight some prominent research questions\nthat BIKED can help address. Of these questions, we further explore the\nfollowing in detail: 1) Are there prominent gaps in the current bicycle market\nand design space? We explore the design space using unsupervised dimensionality\nreduction methods. 2) How does one identify the class of a bicycle and what\nfactors play a key role in defining it? We address the bicycle classification\ntask by training a multitude of classifiers using different forms of design\ndata and identifying parameters of particular significance through\npermutation-based interpretability analysis. 3) How does one synthesize new\nbicycles using different representation methods? We consider numerous machine\nlearning methods to generate new bicycle models as well as interpolate between\nand extrapolate from existing models using Variational Autoencoders. The\ndataset and code are available at http://decode.mit.edu/projects/biked/.\n