Clustering is widely used in the context of the text, especially in classification. On the other hand, text pre-processing has a major impact on improving accuracy. This study aims to study of effect text pre-processing on improving the accuracy of hadith text, and building a model to classify the hadith categories into Saying, Doing, Reporting, and Describing, according to what was attributed to the Prophet Muhammad (PBUH), using learning algorithms. Also, this study is concerned with studying the impact on improving the accuracy of classification on text on different classifications of Hadith according to the text of Hadith and four categories of accuracy.  Two Way Cluster Analysis was used to classify Hadiths according to the tags present in each Hadith. Results showed that the number of tags could be used to distinguish different classes of Hadith especially the Da'if and Maudu as the classification results showed distinct groups of both types while classification showed low accuracy or capability of the separation between Sahih and Hasan Hadiths. Also, the experimental results showed that the best classifier had given high accuracy was SGD Classifier; it achieved higher accuracy followed by Logistic Regression, and finally, Bernoulli NB.