It is difficult for current student behavior analysis systems to simultaneously achieve efficient data processing, convenient client management, and real-time response to large-scale data. This paper constructed an efficient student behavior analysis system that adapted to large-scale data by optimizing the B/S architecture (Browser/Server Architecture) and improving data processing performance. In the system design, MySQL database index optimization, distributed cache Redis, asynchronous data transmission technology and data analysis algorithm based on K-means clustering are used. The query delay is reduced through database optimization, the Redis cache technology improves the concurrent processing capability, the asynchronous transmission reduces the network delay, and the K-means algorithm is used to enhance the depth of data analysis. The experimental results show that the average response time of the B/S architecture under high concurrency is 620 ms, and the throughput reaches 1000 requests/second, which is better than the 790 ms and 800 requests/second of the C/S (Client/Server) architecture. In the data transmission efficiency test, the B/S architecture has a delay of 240 ms in a low-quality network, while the C/S architecture has a delay of 300 ms. In the data processing accuracy verification, the K-means silhouette coefficient of the B/S architecture is 0.73, higher than the 0.64 of the C/S architecture, and the confidence of association rule mining is 92%, higher than the 85% of the C/S architecture. The results show that the student behavior analysis system based on B/S architecture has stronger performance and accuracy when dealing with large-scale data.