Random forest sklearn cross_val_score anomaly

Does scikit-learn cross_val_score function normally allows you to predict the score of your submission? I am facing a weird behaviour where I am getting cross_val_score value of ~0.81, but my submission earns only 0.68. Is this normal behaviour, can this even be explained? I am using RandomForestClassifier, features are: Age, Pclass, family size, Sex. Any insights would be greatly appreciated. Rokas


Link to Full Article: Random forest sklearn cross_val_score anomaly