In this paper, the application of intelligent algorithms such as XGBoost and Char-CNN is verified by establishing the analysis model of the game Wordle. The three models established can perform the tasks of fitting and predicting the number of players, analyzing the number of attempts, and classifying the difficulty of the game. In the first model, XGBoost optimizes the classical SIR Model to improve output accuracy and analysis efficiency. In the second model, this paper designs an innovative two-dimensional code word coding method concerning machine language, which effectively reduces the redundancy of Char-CNN. In the third model, the classical Bayesian algorithm is used to classify the difficulty of a large word set, proving its usability. In addition to helping Wordle editors identify more appealing puzzles, the process of analyzing text with intelligent algorithms is also valuable for scholars who study the text features. In the era of big data, the model-building method in this paper may also give some enlightenment to the field of artificial intelligence.