Funded by the European Union. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European Education and Culture Executive Agency (EACEA). Neither the European Union nor EACEA can be held responsible for them.


A Study of Cross-Validation and Bootstrap for Accuracy Estimation and Model Selection

Users: 1 - Average Rating: 5.00

This paper reviews accuracy estimation methods and compare the two most exploited techniques: cross-validation and bootstrap.
The authors explain some of the assumptions made by the estimation methods, and present concrete examples where each method fails. The authors, in particular, are interested in identifying a method that is well suited for the biases and trends in typical real-world datasets.
The results obtained by the authors in this paper indicate that for the considered dataset and similar ones, the best method to use for model selection is ten-fold stratied cross validation, even if computation power allows using more folds.

Scientific Paper

Data Analytics, Machine Learning

Target Group:

Not available.

Cite as:
Kohavi, R., A study of cross-validation and bootstrap for accuracy estimation and model selection, Ijcai 14.2 (1995): 1137-1145.

Author of the review:
Giulia Cademartori
University of Genoa


You have to login to leave a comment. If you are not registered click here

Joana Lopes

Useful for my Master research on the topic «cross-validation».