This paper provides a survey and practical perspective on the application of machine learning (ML) in industrial sensing and control. It identifies key statistical and ML techniques that have achieved practical success in the process industries. The paper discusses hybrid modeling as a methodological framework for core application areas: soft sensing, process optimization, and control. Soft sensing involves the use of statistical and ML methods for estimating variables that are difficult to measure directly. The paper quantifies research trends and highlights the most successful techniques in practice. It considers two distinct approaches for data-driven optimization and control: hybrid modeling combined with mathematical programming techniques and reinforcement learning. The paper discusses the challenges of purely data-driven methods, emphasizing the need to balance deep learning with domain knowledge. It also highlights ways to integrate prior knowledge into industrial ML applications. The paper reviews the history and current state of ML in the process industries, noting the long-standing success of statistical techniques and the growing interest in more recent ML architectures. It discusses the practical deployment of ML techniques in the process industries, including the use of hybrid modeling, which combines knowledge-driven and data-driven models. The paper also addresses the computational cost of soft sensors and the challenges in their development, such as the lack of labeled data and the need for maintenance. It explores data-driven and hybrid modeling approaches for optimization and control, including model-based optimization, model predictive control, and reinforcement learning. The paper highlights the potential of reinforcement learning in process control, noting its ability to handle complex, nonlinear systems and its adaptability to changing conditions. The paper concludes with a discussion of the challenges and opportunities in applying ML techniques to industrial processes.This paper provides a survey and practical perspective on the application of machine learning (ML) in industrial sensing and control. It identifies key statistical and ML techniques that have achieved practical success in the process industries. The paper discusses hybrid modeling as a methodological framework for core application areas: soft sensing, process optimization, and control. Soft sensing involves the use of statistical and ML methods for estimating variables that are difficult to measure directly. The paper quantifies research trends and highlights the most successful techniques in practice. It considers two distinct approaches for data-driven optimization and control: hybrid modeling combined with mathematical programming techniques and reinforcement learning. The paper discusses the challenges of purely data-driven methods, emphasizing the need to balance deep learning with domain knowledge. It also highlights ways to integrate prior knowledge into industrial ML applications. The paper reviews the history and current state of ML in the process industries, noting the long-standing success of statistical techniques and the growing interest in more recent ML architectures. It discusses the practical deployment of ML techniques in the process industries, including the use of hybrid modeling, which combines knowledge-driven and data-driven models. The paper also addresses the computational cost of soft sensors and the challenges in their development, such as the lack of labeled data and the need for maintenance. It explores data-driven and hybrid modeling approaches for optimization and control, including model-based optimization, model predictive control, and reinforcement learning. The paper highlights the potential of reinforcement learning in process control, noting its ability to handle complex, nonlinear systems and its adaptability to changing conditions. The paper concludes with a discussion of the challenges and opportunities in applying ML techniques to industrial processes.