Topic analysis algorithm is a commonly used data mining technology, which can mine potential topics and the correlation of these topics from text data. While, the traditional topic analysis algorithm has shortcomings in efficiency and accuracy when dealing with large-scale data. To solve the above problems, this paper designs a new data analysis algorithm, which combines the topic analysis model and T-test method. Through the processing and verification of Japanese teaching related text data, the algorithm in this paper is effective. First, the topic analysis is used to model the text data and extract the potential topic information. Then, the T-test method is used to measure the significance difference between different topics and screen out the topics with significant impact. By combining probabilistic models with statistical methods, the algorithm can find and analyze key topics in data more accurately, and improve the robustness of topic analysis. Data experiments show that the data analysis algorithm in this paper performs well in extracting potential topics and analyzing topic differences.