With the advent of the big data era, data analytics is playing an increasingly important role in discovering valuable information from massive data and guiding users to make better decisions. However, there are three major challenges in the data analytics process: high coupling of analytics workflows, multiple types of interaction interfaces, and high time consumption for exploratory analytics. To address these challenges, we propose Navi, a data analytics system based on natural language interaction. Using modular design principles, Navi abstracts three core functional modules from mainstream data analytics workflows: data query, visualization generation, and visualization exploration, reducing the coupling of system design. Moreover, Navi uses natural language as a unified interface for interaction and enables efficient coordination of various functional modules through a task scheduler. Furthermore, to solve the problems of exponential search space and unclear user intent in visualization exploration, we propose a visualization automatic exploration method based on Monte Carlo tree search and design a pruning algorithm and a composite reward function based on visualization domain knowledge, which improve the search efficiency and result quality. Finally, we validate the effectiveness of the Navi through quantitative experiments and user studies.