Cross-Validation Explained

Cross-validation is a technique that tests a model's performance more reliably by splitting the data into multiple parts and training and testing the model several times, instead of just once.

Why cross-validation matters

• A single train-test split can give misleading results by chance.
• It gives a more accurate, stable measure of model performance.
• It makes better use of limited data, since every part gets used for both training and testing.

K-Fold Cross Validation

The dataset is divided into K equal parts, called folds. The model trains on K minus 1 folds and tests on the remaining one, repeating this process until every fold has been used for testing.

from sklearn.model_selection import cross_val_score

from sklearn.linear_model import LogisticRegression


model = LogisticRegression()

scores = cross_val_score(model, X, y, cv=5)


print(scores)

print(scores.mean())

cv=5 means the data is split into 5 folds, so the model is trained and tested 5 times. scores.mean() gives the average accuracy across all folds, which is a more trustworthy measure than a single test result.

Cross-validation helps confirm that a model's performance is consistent, and not just a result of a lucky data split.