This research looks at different ways computers can automatically spot skin cancer from images of skin problems. They tested two main approaches. The first one pulls out important details from images using basic tools and then feeds those details into various machine learning programs to make a diagnosis. The second approach uses more advanced deep learning methods like convolutional neural networks that can both find the important image details and make the diagnosis all in one go. The researchers tried out several traditional machine learning methods including Logistic Regression, k-Nearest Neighbors, Naive Bayes, Decision Tree, Random Forest, and Support Vector Machine. They also tested four different CNN models: VGG16, ResNet60, InceptionV3, and Inception-ResNetV2. For the first approach, they used four different ways to pull out image features and combined them together to get the best possible feature set, which should lead to better accuracy in spotting cancer. To test how well these methods work, they used a collection of skin lesion images from the ISIC database that includes both harmless and cancerous skin conditions. They measured how effective each algorithm was using standard performance metrics and also kept track of how long each one took to train, giving them insight into both accuracy and computational demands.