Thank you for the interesting presentation. I have some questions regarding the method applicability and the evaluation slide (slide number 6): How practical is your method in actual use cases compared to the simulation environment? How do you see your approach being adopted in the industry? How does this approach perform for more complex scenarios (e.g., with more jobs) – is the method scalable? What is the measurement unit of the table from slide 6? What does it change from one test instance to the other? Why were the trained models LA01 and LA16 selected for the table? Thank you; I look forward to the answers.
Thank you for your question. I will do my best to answer them for you.
1. Current practicability: The simulation is a simplification, no transition times (e.g. transport times) or set-up times (possibly depending on the product change) are considered.
2. Adoption in industry: I see our approach as complementary to currently used operations research methods. Thus, common planning systems (APS systems) can be extended.
3. In our paper we only compared two complexities (explicitly). In our research (implicitly) we have already obtained good results on even more complex scenarios. Which means that the approach scales well, although we speak of “generalization” rather than “scaling”.
4. Slide 6 units: These are abstract time steps.
5. Slide 6 table: Each test instance has its own optimal lead time. I admit, this makes the comparison between the instances a bit more difficult and a normalized size would have been more helpful here for a good comprehensibility.
6. Selection of LA01 and LA16: These instances were chosen because they were the first instance in their complexity group. Here LA01-LA05 form a complexity group, as do LA06-LA10 and LA11-LA15 and LA16-LA20. We chose these two complexity groups because they are slightly further apart and thus more divergent from each other. Thus, we forced the generalization check of our approach.
I hope this answer has adequately answered your questions. If not, feel free to contact me directly via email.
Thank you for the interesting presentation. I have some questions regarding the method applicability and the evaluation slide (slide number 6): How practical is your method in actual use cases compared to the simulation environment? How do you see your approach being adopted in the industry? How does this approach perform for more complex scenarios (e.g., with more jobs) – is the method scalable? What is the measurement unit of the table from slide 6? What does it change from one test instance to the other? Why were the trained models LA01 and LA16 selected for the table? Thank you; I look forward to the answers.
Thank you for your question. I will do my best to answer them for you.
1. Current practicability: The simulation is a simplification, no transition times (e.g. transport times) or set-up times (possibly depending on the product change) are considered.
2. Adoption in industry: I see our approach as complementary to currently used operations research methods. Thus, common planning systems (APS systems) can be extended.
3. In our paper we only compared two complexities (explicitly). In our research (implicitly) we have already obtained good results on even more complex scenarios. Which means that the approach scales well, although we speak of “generalization” rather than “scaling”.
4. Slide 6 units: These are abstract time steps.
5. Slide 6 table: Each test instance has its own optimal lead time. I admit, this makes the comparison between the instances a bit more difficult and a normalized size would have been more helpful here for a good comprehensibility.
6. Selection of LA01 and LA16: These instances were chosen because they were the first instance in their complexity group. Here LA01-LA05 form a complexity group, as do LA06-LA10 and LA11-LA15 and LA16-LA20. We chose these two complexity groups because they are slightly further apart and thus more divergent from each other. Thus, we forced the generalization check of our approach.
I hope this answer has adequately answered your questions. If not, feel free to contact me directly via email.
Best
Till Saßmannshausen