Reviews: A Meta-Analysis of Overfitting in Machine Learning

Update: I thank the authors for the feedback and the additional figure. I would gladly vote for acceptance of the paper. They surveyed a wide range of 112 Kaggle competitions and concluded that there is little evidence of substantial overfitting. I found the question very important to the machine learning community. The investigation is thorough that it covered a broad spectrum of datasets.