Splitting the data

HR-аналитика: прогнозирование текучести кадров на Python

Hrant Davtyan

Assistant Professor of Data Science American University of Armenia

Target and features

target = churn
features = everything else

Train/test split

train - the component used to develop the model
test - the component used to validate the model

from sklearn.model_selection import train_test_split

target_train, target_test, features_train, features_test =
                           train_test_split(target,features,test_size=0.25)

Overfitting

an error that occurs when model works well enough for the dataset it was developed on (train) but is not useful outside of it (test)

Let's practice!

HR-аналитика: прогнозирование текучести кадров на Python