A VANET was simulated with malicious nodes introduced in the DSR file. After a simulation of the network Packet Trace data and Application Metric data was obtained. The data underwent data preprocessing with pandas and numpy library.Feature selection was done with an algorithm called RECURSIVE FEATURE ELIMINATION and then machine learning models like Logistic Regression,K-Nearest Neighbour, Decision Tree, Feed Forward Neural Network and Random forest Classifier were made and trained on the data...On test and validation it was found tree based algorithms perform best at detecting sinkhole attack with an impressive accuracy and mean precision.