Data science has emerged as a new paradigm and is fundamentally changing the way we approach in all areas of science and engineering. In contrast with traditional hypothesis-based approaches, unbiased data analyses can provide new insights and discoveries that could not be obtainable in the past. With increasing availability of big data and advanced digital technology, data-driven modeling serves as a key driver of this scientific revolution. Data-driven modeling is particularly useful for predicting complex human-like behaviors of microorganisms such as decision making, anticipation, adaptation, communication, and networking. In the first part of the seminar, I will compare reinforcement learning with the cybernetic approach to understand their connection and differences in modeling microbial intelligence. The second part will include big data-based supervised learning, e.g., to infer spatiotemporal microbial interactions from image data vs. sparse identification of model equations from limited data, e.g., to predict microbial inactivation processes in food. These demonstrations offer the data-driven modeling as an ideal framework to build digital twins of complex biological systems for detection, diagnosis, prognosis, control, and design.