ANAMOLY DETECTION

This project uses machine learning to detect Neptune attacks, a type of denial-of-service attack, in network activity data. We trained several models, including Logistic Regression, K-Nearest Neighbors, Random Forest, Support Vector Classifier, and Decision Tree, on a dataset of 86,845 records. The best-performing model was Random Forest, which achieved the highest accuracy. The dataset includes train.csv with training observations and target values, and test.csv with testing observations without target values. Using the trained model, we predicted the target values for 21,712 test set entries. The predictions were formatted according to the sample-submission.csv file and saved for submission, with 'attack' as the column name. The target column indicates normal activity (0) or a Neptune attack (1).