The current rapid growth in the computer and internet development has ushered in numerous cybersecurity challenges which are constantly evolving with time. The current cybersecurity solutions are no longer optimal in tackling these emerging cyber threats and attacks. This paper proposes the creation of a cybersecurity dataset to be used for a hybrid machine learning (ML) approach of supervised and unsupervised learning for an effective intrusion detection system. The proposed model entails a five-stage process which starts at the setup of a simulated network environment of network attacks to generate a dataset which feeds into the data normalization stage and then to data dimension reduction stage using the principal component analysis as a feature extraction method after which the data of reduced dimension is clustered using the k-Means method to bring about a new data set with fewer features. This new dataset is afterward classified using the enhanced support vector machine (ESVM). The proposed model is expected to provide a high-quality dataset and an efficient intrusion detection system in terms of intrusion detection accuracy of 99.5%, short train time of 3seconds and a low false-positive rate of 0.4%.