The liver is the most significant organ in the human body since it handles a significant role in food digestion and progression in our body. Mainly it takes an essential part in enzyme activation, fat metabolism, bile synthesis, vitamin, glycogen, and mineral storage. Depending on the role it controls, it has a sophisticated accidental of coming in contact with harmful creation that goes inside the body. Hence, the diagnosis of liver disorder has been subjective at best, based on subjective approaches. Liver disorders are challenging to detect, and as a result, they are regularly overlooked in the early stages due to a lack of precise symptoms. Hyperbilirubinemia is one of the most substantial signs of most liver illnesses, and it can be demanding to differentiate early on. However, in most of cases, this isn't certain, and the ability to detect and confirm the presence of liver disease lead to a better understanding of enzyme levels. The prediction of liver illnesses has been done using a variety of machine learning techniques. In this investigation, the recommended ensemble soft voting classifier offers binary classification and utilize the ensemble of three machine learning algorithms: Decision Tree, Support Vector Machine, and Naive bayes classifiers to predict and expose liver disease by Binary Classification of the dataset into two particular types of patients with or without liver disease (patient suffering liver sickness or not). The unbalanced dataset comprises materials about human patient attributes such as Gender, Age, Alanine, Total Bilirubin, Aminotransferase, Aspartate Aminotransferase, Direct Bilirubin, Albumin, Alkaline Phosphatase, Globulin Ratio and the Result and Total Proteins Albumin. Furthermore, the accuracy and various error calculations of the predictions from the aforementioned algorithms are analyzed to recognize and identify the best- convenient algorithm.