The Covid-19 pandemic has greatly affected the world, including Malaysia. Analysis of the spread of Covid-19 using epidemiological methods is useful for researchers in estimating how the virus might spread, as well as for health agencies in planning effective measures to prevent transmission in the future. Machine learning (ML) techniques have been widely adopted in analyzing and predicting various aspects of the pandemic. In this study, four ML techniques, artificial neural network (ANN), support vector regression (SVR), random forest (RF), and gradient tree boosting (GTB), were applied to predict new cases of Covid-19 in Malaysia. The study aimed to identify which ML model provided the best prediction accuracy for the Malaysian Covid-19 dataset. Experimental findings showed that GTB had the smallest quantitative error values and provided the best predictive capability for new Covid-19 cases. SVR and ANN followed behind, while RF performed the worst. The study concluded that ML models have promising predictive capabilities in analyzing Malaysia's Covid-19 data, and future improvements such as optimization of model parameters are recommended to improve predictive abilities.