Assessing the impact of Principal Component Analysis on accurately predicting melanoma diagnosis applied on different classification models
View/ Open
Date
2019-11-26Author
Olmedo Rivera, Juan Cristobal
0000-0002-4664-0586
Metadata
Show full item recordAbstract
With huge amounts of data at our disposal in the medical field, mathematical models are built to diagnose diseases. This study focuses on melanoma because it’s the type of skin cancer that accounts for most deaths, up to 7,230 in 2019 according to the American Cancer Society. The study focuses on the effectiveness on diagnosing melanoma and how Principal Component Analysis (PCA) impacts the performance of four models being assessed, which are: K Nearest Neighbor (KNN), Logistic Regression (LR), Support Vector Machines (SVM), and Artificial Neural Networks (ANN). Each model evaluates the melanoma dataset before and after performing the PCA transformation. Results show that PCA does not impact performance in this case. Even though PCA does not improve performance, the modeled results achieve better results when compared to dermatologist and other algorithms.