With the advancing technology, the increase in threats has been exponential. These technologies have led to the production of huge amounts of network traffic data. Therefore, it is of immense importance for the companies to safeguard this sensitive data from any attack by the implementation of robust security techniques as this can cause heavy financial losses to the companies. There is need for the enhancement in the performance of existing Intrusion Detection Systems in order to detect the attacks more diligently. This is done by the integration of Machine Learning in the intrusion detection approaches as they develop models which are data driven and hence detect and prevent cyber-attacks. This study develops an ML based intrusion detection model by using Random Forest, Naïve Bayes and XGBoost classifiers on UNSW_NB15 dataset and pre-processing the dataset by applying One Hot Encoder and thereby giving an accuracy of 99.98%, 74.59% and 66.20% respectively. The performance evaluation indicates that the Random Forest model with an accuracy of 99.98% clearly outperforms the other algorithms. Hence, this framework can improve the system security to a greater extent.