Speaker
Description
This project studies the use of random forest algorithm in discriminating jets initiated by quark or gluons. Jets are collimated flow of hadrons from quarks and gluons produced in collisions clustered using dedicated algorithm. The random forest which is an ensemble decision tree is discussed in detail on why it is suitable to this kind of data. Monte Carlo generated dataset is used to train the random forest algorithm and the performance is compared to the gradient boosted decision tree and multilayer perceptron neural network which are predominantly used in the ?previous literature. From this project, we see that the random forest performed slightly better than the other two algorithms with an accuracy of 76.66%. This may be due to it's ability to train on the whole dataset and use out-of-bag data to test performance of the algorithm.