Leveraging a Heterogeneous Ensemble Learning for Outcome-Based Predictive Monitoring Using Business Process Event Logs

Round 1

Reviewer 1 Report

This paper takes advantage of stacking ensemble method for outcome-base predictive monitoring. In general, the paper is well written and easy to follow. The results are robust as they tested the method in multiple datasets and with multiple evaluation metrics. My main concern is what is the “advanced heterogeneous ensemble learning” in the title. Stacking is in general not new which has already been used widely in practice. It is not clear what innovations are made on the general stacking method in this research. Another concern is the multiple folds cross validation method may not fit for time series data, as it may suffer the problem of information leak when using future data as the training sets to predict past.

Author Response

See attached file

Author Response File: Author Response.pdf

Reviewer 2 Report

This is a carefully done study and the findings are considerable interest. It contributes a advanced stacking ensemble technique for outcome-based predictive monitoring, which enables improve the performance of predictive models. However, there are a few minor issues that require further revision, and some conclusions need to be supplemented. My detailed review is as follows:

1.     There are too many expressions about the achievements of others in the second part(line 79), it is recommended to delete the unnecessary part, and the remaining part is integrated with the content of the first section.

2.     In Figure 1 in the result display, the results of the same attribute obtained by different methods are represented by the same polyline, which is somewhat inappropriate. Maybe the way it's presented can be improved a bit more.

3.     The proposed method uses three strong learners to form a stacked ensemble. Is the presence of all three strong learners necessary? Have ablation experiments been considered? Maybe this part can be improved.

Author Response

See attached file

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

I understand that the authors have made much effort on this paper, the only issue is its lacking of innovations. It is not a surprise to me that stacking works in this application, but I agree this could be an application case.