Impure leaf nodes in decision tree made using gini impurity

This tree has been built using gini impurity now during evaluation if we reached at left most node of tree which has value 17/3
it means 17 rows of dataset says YES,3 rows of dataset says NO,
so what should be my final output should i go with majority??Is this how scikit learn works in case of impure leaf nodes??

© Copyright 2013-2019 Analytics Vidhya