CV and data Split

M1. The Predictive Modeling Pipeline M1. Model evaluation using cross-validation
#1

I would like to understand the difference between the data split between train and test e CV.
It is substitute techniques or it is complementary techniques?

#2

I think this is explained and motivated in the “Validation of a model” video.

A train test split gives a single number to score the model. Cross-validation gives several numbers (each number corresponding to the score on a particular fold) that give a confidence interval around the mean score of the model.

1 Like