Breast cancer detection is the initial stage of a cancer diagnosis. Therefore, more accurate classifiers are always preferred. A high accuracy classifier offers an extremely low probability of incorrectly classifying a cancer patient. This research investigates the performance of a modified and improved version of the guiding hypothesis of the logistic regression. Gradient descent and complex optimization techniques are both used to minimise the cost function. The weighting factor for the hypothesis, which is a sigmoid function, is determined by the number of features, the size of the dataset, and the kind of optimization technique used, as can be seen. By carefully choosing the value, which relies on the number of features and the kind of optimization methods utilised, the accuracy of breast cancer diagnosis is significantly increased. "The accuracy, sensitivity, and specificity of the findings improved considerably, which was a good sign.'' The worldwide danger that breast cancer presents to women's health makes it particularly important to distinguish benign from malignant tumours based on ultrasound images. Despite the fact that both morphological and texture features are necessary for accurately representing ultrasound breast tumour images, their simple combination has little effect on the classification of benign and malignant tumours because high-dimensional texture features are too aggressive and obscure the significance of low-dimensional morphological features. An efficient textural and morphological feature combination strategy is suggested to more accurately differentiate benign and cancerous tissue. First, elements from the morphological (such as form complexity) and texture (such as local binary patterns [LBP], histograms of oriented gradients [HOG], and gray-level cooccurrence matrix [GLCM]) components of breast ultrasound images are retrieved. A support vector machine (SVM) classifier operating on texture features is trained, and a Navie Bayes (NB) classifier is constructed in order to take use of the discriminative potential of texture features and morphological traits, respectively. Thirdly, the combined weighted classification scores from the two classifiers are used to construct the final classification result (SVM and NB). The low-dimensional nonparameterized NB classifier efficiently handles the parameter complexity of the overall classification system when paired with the high-dimensional parametric SVM classifier. Thus, morphological and textural components are successfully combined. The recommended technique outperforms several comparable benign and malignant breast tumour classification methods with a 91.11% accuracy, 94.34% sensitivity, and 86.49% specificity, according to thorough experimental evaluations that are reported.