Heart failure stands as a significant public health issue with the high mortality and morbidity rates. Timely anticipation and detection of heart failure play a vital role in facilitating prompt intervention and improved prognoses. Recently, machine learning methodologies have emerged as auspicious tools. Machine learning approaches have emerged as promising tools for predicting and diagnosing heart failure in prognosticating and identifying heart failure cases. This survey paper presents an extensive overview of the extant body of literature concerning machine learning cantered strategies for the early prognosis and diagnosis of heart failure. We scout into exploration of diverse ML techniques that are frequently employed in heart failure prognosis and diagnosis, including artificial neutral network, random forest logistic regression, support vector machine models and deep learning. Furthermore, an examination of the datasets and evaluation metrics used to calculate the efficiency of machine learning models in prognosis and diagnosing heart failure is presented. Our investigation culminates in a synthesis of the principle discoveries compiled from the reviewed literature. We undertake a comparative analysis of performance exhibited by distinct ML technologies, while also addressing the obstacles and constraint inherent in that the utilization of machine learning for heart failure prognosis and diagnosis. By offering this survey article, we furnish valuable asset of researchers and medical practitioners who possess a wasted interest in elevating machine learning techniques for the early prognosis and diagnosis of heart failure.