Prediction of breast cancer using Random Forest Classification on the Wisconsin Breast Cancer Dataset. Implemented with Streamlit.
Histological diagnosis of breast cancer is the gold standard, but is time-intensive. By enabling pathologists to automate analysis of cell slides, we can provide an effective screening tool to analyse slides, and in many cases, provide professional support in countries which lack the resources to train more pathologists.
This dataset was trained with SKLearn and implemented with Streamlit to provide the front-end web app.
4 models are implemented to switch between for testing: Random Forest Classifier, Naive Bayesian Classifier, K Nearest Neighbours, and a Decision Tree Classifier.
Random Forest Classification gives the best accuracy (~92%)
This web app is not suitable as a diagnostic tool and is intended for teaching purposes only.