Fig. 3From: Predicting cash holdings using supervised machine learning algorithmsRMSE with different k settings. With all 19 features, the best hyperparameter k is 27, whereas with reduced 15 features, the best k is 57Back to article page