<![CDATA[Dreams are considered important spiritual experiences in Islam. Muhammad Qasim’s dreams, associated with the end times, have generated various public responses. This study aims to explore patterns of public response to these dreams using a data mining approach. Data were collected through an online questionnaire covering six main variables: dream experiences, knowledge of dreams in Islam, belief in the Apocalypse, knowledge of end-time science, and knowledge about Muhammad Qasim’s dreams, with responses from 708 participants used as class labels (positive, neutral, negative). The methods employed include CHAID for classification, and FP-Growth for association rule mining. The results show that CHAID produces a compact and interpretable model, generating 9 rules with an accuracy of 57.63%. The precision values for the negative, neutral, and positive classes are 30.77%, 38.46%, and 62.28%, while the recall values are 7.55%, 21.05%, and 87.38%. The F1-scores are 12.14%, 27.20%, and 72.72%, with a macro F1-score of 37.35% and a weighted F1-score of 64.04%. The hybrid CHAID–FP-Growth approach produces a higher proportion of high-quality rules based on support, confidence, and lift compared to conventional FP-Growth. In addition, an exact matching rule is identified across both methods: Knowledge about Muhammad Qasim’s dreams = Yes → Response = Positive, with a support of 26.5%, confidence of 83.2%, and lift of 1.436. This study contributes by proposing a segment-based hybrid approach that enhances the quality of association rules and identifies cross-validated patterns across methods. Overall, the model demonstrates fairly good performance, but still has limitations in handling minority classes.]]>
