The dreadful condition known as neoplastic breast cancer seriously endangers the health of women. It is considered to be the greatest contributor to female malignant development-related deaths. To reduce the mortality rate from bosom sickness, accurate diagnostic confirmation and efficient treatment are essential. Strong illness disclosure has lately been a popular application for machine learning )ML( techniques, with irregular woods being one of the most often used. In any case, it is conceivable for the preparation cycle to result in decision trees with high similarity and poor grouping execution, which might have a detrimental effect on the model's overall arrangement execution. A Hierarchical Clustering Random Forest )HCRF( model has been created as a result of all of this effort. To assess the proximity of all the trees, a unique tiered bunching approach is employed to group the selected trees. To build the different levels bunching arbitrary backwoods with high precision and little resemblance, test trees are compared to isolated groups. Moreover, we use the Variable Importance Measure VIM approach to increase the chosen inclusion number for the prediction of bosom malignant development .Both the Wisconsin Breast Cancer data set from the UCI College of California, Irvine AI vault and the Wisconsin Diagnosis Breast Cancer WDBC data bank were used in this study. The presentation of the recommended method is evaluated in terms of its accuracy, correctness, responsiveness, explicitness, and AUC. The outcomes of the trials on the WDBC and WBC datasets demonstrate that the classification based on the HCRF calculation employing VIM as a component determination approach achieves the highest exactness, with 97.05% points and 97.76%, respectively Compared to Decision Trees, Ada boost, and Irregular Woods the approaches used in this study might be used to diagnose bosom disease