IMPORTANT : The culmination of our work is in ProjetFinal.ipynb
Recently, AI and machine learning have taken on a major importance in the field of health. Indeed, the possibility of automatically carrying out diagnoses has made our health system (contrary to appearances) much more efficient. In this context, we wanted to improve our understanding of supervised classification by studying the case of classification of benign versus malignant tumors.
To do this, we tried to experiment with different models that we have made entirely or almost entirely from scratch. We have tried to graphically visualize our data in order to better understand it and be able to better derive them. We finally focused on 3 models: a naive Bayesian classifier, a generalized linear model (GLM) and the k-nearest neighbor (KNN) method which seemed to be used in several of the most powerful models that exist to date in the field of tumor classification.