Experimental Analysis on Breast Cancer Using Random Forest Classifier on Histopathological Images

The rising female death rate can be directly attributed to the prevalence of breast cancer. Due to the lack of resources and the lengthy time required for human diagnosis, an automated diagnosis system for cancer screening is desperately needed. Predicting breast cancer using machine learning might reveal previously unknown traits. In addition, manual detection is arduous work that increases the likelihood of making mistakes and opens the door to misclassification on the part of pathologists. This study introduces a Random Forest Model for the automated identification of breast cancer based on histopathology pictures from the popular Breast Cancer Kaggle dataset, which aims to overcome the aforementioned problems. The study's overarching goal is to learn what characteristics best predict which types of cancer will develop, as well as to spot broader patterns that might guide future model and hyper parameter choices. The aim is to determine the aggressiveness of the breast cancer. A function that can anticipate the incoming input's discrete class has been fitted using machine learning classification methods to do this. Newly generated pictures are used to evaluate the proposed work, with the results being better than those of previous classifiers. Results show that Random Forest Classifier, with an accuracy of 95.67 percent and a reduced error rate, performs better than Deep Convolutional Neural Network.