I have used the concept of ensemble models once using Max Voting and then Rank Average.
I have ensembled three classifier models: Logistic Regression, KNearestNeighbors and Decision Tree Classifier.
KNearestNeighbors has been checked dor Brute Force, kd-tree and ball-tree algorithms.
Decision Tree Classifier has been checked for quality by both gini impurity and information gain. The split for nodes has been checked by both best split method and random best split method.