Análisis comparativo de predicción dentro de bases de datos de cáncer: Una aplicación de aprendizaje automático

Cancer is not only a disease, it is a set of diseases with a great impact on public health. In that sense, efforts to consolidate methods of analysis based on large data that contribute to its prediction, is an area of special interest for scientists and data analysts. The objective of this paper is to compare the performance of method prediction: i) Logistic regression, ii) K Nearest Neighbor, iii) K-means, iv) Random Forest, v) Support Vector Machine, vi) Linear Discriminant Analysis, vii) Gaussian Naive Bayes viii) Multilayer Perceptron, within a cancer database. In the case of unsupervised learning models, the relevance of the centroids for the k means algorithm is evident, as well as the learning rate assignments and parameters for the Multilayer Perceptron case. In the case of supervised learning models, SVM performs best.