The novel coronavirus causing COVID-19 ailment has wreaked havoc worldwide with unprecedented causalities and massive losses. There have been various trials to contain the spread of the disease with early symptomatic treatment, disease estimates, smart lockdowns etc. However, one of the most effective measure to minimize casualties have been the early and accurate detection of the onset of the disease. Incidentally, with the sudden upsurge of the cases, medical and diagnostic procedures have seen severe constrains both in terms of scans and radiologists studying the scans to pass their verdict regarding the severity of the diseases and further possible progress. This has led to researchers in exploring automated techniques which would ease certain amount of pressure on the medical system for accurate and timely diagnosis. With the copious amounts of data to be processed and the diversity in the cases, machine learning happens to be the front line contender among statistical techniques. Several machine learning based techniques have been developed and are also being developed with new strains of the virus emerging through mutations worldwide. This paper presents a comprehensive review on the background of the novel coronavirus and the need for brisk and accurate automated detection of the disease, the role of machine learning for the same and various machine learning based approaches adopted thus far in contemporary literature for the automated detection of COVID1-19. The paper highlights the salient points of the existing techniques with palpable limitations of existing techniques which can served as a stepping stone for the design and implementation of advanced techniques which would beat existing approaches in terms of classification accuracy.