Checking if ML model is possible

Question

How can I check if a machine learning model is feasible on a given dataset? What techniques like EDA, correlation etc. can be used to judge if a model is possible i.e. data and predictor variables will give reasonably accurate forecasts or in other words there is good enough signal in predictors?