PyProgressive: A Lightweight Library for Progressive Computations

Abstract

Progressive computation has emerged as a promising paradigm for handling large datasets by providing early, intermediate results within bounded response times. However, transforming existing sequential computations into their progressive counterparts often requires substantial manual effort and algorithmic restructuring. To address this gap, we present PyProgressive, a Python library for specifying and executing progressive computations. With PyProgressive, users can express their computations using a simple set of sequential-style primitives, while the system automatically compiles these primitives into a form that can be updated and evaluated progressively. We demonstrate the utility of PyProgressive through statistical and visualization tasks, including the progressive group-wise proportions.