Wordle, an engaging online crossword puzzle, challenges players to decipher a hidden five-letter word. In this manuscript, our primary focus lies in predicting the distribution of reported results in the Wordle game. To accomplish this, we devised and compared multiple predictive models, seeking the most accurate and stable solution. The foundation of our research revolves around training 31 distinct word attributes through five diverse multi-output regression data processing models. Then we choose a model whose average error of the test set error is very small, indicating that the prediction accuracy of our model is better, and also more stable than other models. Finally, to illustrate the practical application of our model, we provide a concrete example of predicting the associated percentages of (1, 2, 3, 4, 5, 6, X) for the word “EERIE” in future game instances.