With much of the easy-to-produce oil already recovered from reservoirs, design and selection of practical Enhanced Oil Recovery (EOR) methods have become increasingly significant in field development planning. This study presents a novel EOR screening workflow using Machine Learning (ML) to predict the optimal EOR strategy for a target reservoir. Data pertaining to six common EOR methods are collected and tabulated to create a database. Various ML algorithms are ranked by performance on screening data. An ensemble model is created using the weighted-average of four algorithms. Feature Selection using Analysis of Variance (ANOVA) method indicates that out of the 22 features initially considered, only 8 have significant impact on the classifier performance. Model refinement is done using hyperparameter tuning and k-fold cross-validation. An increment of 10 percent in accuracy and recall each, 26 percent in precision, and 17 percent in F1 score of classifier is observed on test data post tuning. Despite the limitations imposed by the lack of reservoir and fluid data, the proposed workflow accurately predicts the most suitable EOR technique in over 90% of cases. Feature Ranking studies indicate that parameter ‘average oil viscosity’ is the most influential parameter in the choice of an EOR scheme. This study presents an opportunity to utilize the potential of ML in the oil industry. The novelty of this work is having shown the ability of ensemble method in combining the outputs from individual ML methods, and produce improved results and more accurate predictions than a singular model.