Driven by the rapid development of digital technology, the digital economy has become a major engine for global economic growth. Accurately predicting its life cycle is the difficulty, it will help policy and development strategy formulation. To the best of our knowledge, this paper for the first time, using large-scale data mining technology, explores an algorithm model to predict digital economy trends and put forward a series of prediction means based on machine learning theory with regression analysis. The basic features and current developments of digital economy are considered, including the discussion of classical methods constraints. A big data-based digital economy trend prediction model algorithm is presented with its essential parts of data collecting, preprocessing processing, model building and optimization. The model efficiency is then verified by typical case studies, and the potential uses in policy making, market analysis and industry evolution are analyzed. The results show that the big data-based forecasting methods are effective in increasing the accuracy of digital economic trend analysis, providing a reliable basis for decision-making for government and enterprises.