Cybersecurity analysis using data science and machine learning plays a crucial role in today's digital era, where information security has become imperative for organizations. This study focuses on the intersection between cybersecurity and data science, using machine learning methods and text analytics to understand and strengthen defenses against threats. The importance of TF-IDF (Term-Inverse Document Frequency) is highlighted as a tool to evaluate the relevance of terms in documents and its application in vulnerability classification.The Multinomial Naive Bayes classifier is presented as an efficient tool in text classification, calculating probabilities of belonging to specific classes based on the frequency of terms. The essential formulas used in this algorithm, such as conditional probability and multinomial distribution, are detailed.The KDD (Knowledge Discovery in Databases) methodology guides the process, from datacollection on platforms like Kaggle to data selection, cleaning and transformation. The use of `TfidfVectorizer` facilitates the discretization of text data, and the `GridSearchCV` method optimizes the model's hyperparameters, achieving an accuracy of 97.36%. Finally, the confusion matrix reveals good overall performance, although areas for improvement are identified, especially in the 'High' class.