A User Study on the Effect of Aggregating Explanations for Interpreting Machine Learning Models