Vega-Lite specifications describe visualizations as encoding mappings from data to properties of graphical marks (e.g., points or bars). The Vega-Lite compiler automatically produces visualization components including axes, legends, and scales. It determines default properties of these components based on a set of carefully designed rules. This approach allows Vega-Lite specifications to be concise for quick visualization authoring, while giving user control to override defaults and customize various parts of a visualization. As we also designed Vega-Lite to support data analysis, Vega-Lite supports both data transformations (e.g., aggregation, binning, filtering, sorting) and visual transformations (e.g., stacking and faceting). Moreover, Vega-Lite specifications can be composed into layered and multi-view displays, and made interactive with selections.