Skin Cancer Detection Using the HAM10000 Dataset: A Comparative Study of Machine Learning Models

Skin cancer is a prevalent and potentially life-threatening disease that requires early and accurate detection for effective treatment. Existing skin cancer detection procedures face limitations in terms of accuracy and efficiency. To overcome these limitations, we put forward a novel solution that leverages machine learning models to classify seven types of skin cancers using the HAM10000 dataset. This dataset consists of diverse clinical images, providing a valuable resource for developing robust classification models. Our approach involves preprocessing the HAM10000 dataset to ensure data consistency and enhance model performance. We apply various machine learning models, including KNN, Decision Tree, Random Forest, Ridge Classifier, and SVM, to classify the different types of skin cancers. Through rigorous evaluation, our models demonstrate promising results. Among the models tested, the Random Forest algorithm exhibits superior performance, achieving a validation set accuracy of 95.5% and a test set accuracy of 95.6%. These results highlight the effectiveness of our proposed methodology in accurately classifying skin cancers and surpassing the performance of traditional detection procedures. By utilizing machine learning techniques, our solution offers an improved approach to skin cancer detection, addressing the limitations of existing methods. The integration of Random Forest further enhances accuracy, enabling early and accurate diagnosis, which is crucial for successful treatment outcomes and improved patient care.