The available datasets provided by our previous works on creep life for nickel-based single crystal superalloys were analyzed through supervised machine learning to rank features in terms of their importance for determining creep life. We employed six models, namely Back Propagation Neural Network (BPNN), Gradient Boosting Decision Tree (GBDT), Random Forest (RF), Gaussian Process Regression (GPR), XGBoost, and CatBoost, to predict the creep life. Our investigation showed that the BPNN model with a network structure of “24-7(20)-1” (which consists of 24 input layers, 7 hidden layers, 20 neurons, and 1 output layer) performed better than the other algorithms. Its accuracy is 1.82% higher than that of the second-best CatBoost regression model, with a mean absolute error reduction of 93.07% and a root mean square error reduction of 88.12%.