Tumour malignancy has caused a very high number of women's deaths. Proper hyperparametric ML technique can help to efficiently detect tumours. In this work, various algorithms to analyse the Wisconsin breast cancer data set have been studied. Deep learning with Adam Gradient Descent Learning has been used, which combines the advance features of adaptive gradient and rms propagation. Linear regression, Shallow SoftMax regression, and Deep neural network SoftMax regression are used for the classification of benign and malignant tumour and have calculated the AUC, ROC and accuracy. A unique hyperparametric change is shown in each model to work on the exactness of the model and also to chalk out correlation between different models. It is concluded in the analysis that for the WBC data set the deep neural network SoftMax regression significantly improves the accuracy from other models, and we reached a test accuracy 0f 0.956 and testing accuracy of 1.0.
展开▼
