(Bug: Alignment) best_worst_cost will not consider the MODEL_COST_FUNCTION from parameter

[lvzheqi]

best_worst_cost will not consider the MODEL_COST_FUNCTION from parameter, because PRAM_TRACE_COST_FUNCION is None, it always calls the "construct_standard_cost_function" and uses the default cost.

when the model-move and log-move have different cost, it is problematic by fitness computing.

[Javert899]

Are you considering latest version?

[Javert899]

See commit https://github.com/pm4py/pm4py-source/commit/1669369caefbb28bb9493079c242cba14172edd7

[lvzheqi]

yep, I am not working on the latest version. It seems good for best-worst-cost. But fitness will be problematic when the model-move and log-move have different cost, right?

Javert899 notifications@github.com 于2019年4月15日周一 下午5:38写道：

Are you considering latest version?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/pm4py/pm4py-source/issues/75#issuecomment-483304198, or mute the thread https://github.com/notifications/unsubscribe-auth/ADUOWBC7LCL2S5RDMVHFQ7TPQSON5ANCNFSM4HGBC7ZQ .

[Javert899]

Here it depends on how the fitness value is calculated :)

For this let me know if you think it is wrong and eventually how to fix that!

[lvzheqi]

For me, I think, fitness = 1 - align["cost"]) / ( len(log[index]) MODEL_MOVE_COST + best_worst_cost LOG_MOVE_COST). I'm not so sure... MODEL_MOVE_COST and LOG_MOVE_COST maybe need to extract from parameters

Javert899 notifications@github.com 于2019年4月15日周一 下午5:53写道：

Here it depends on how the fitness value is calculated :)

For this let me know if you think it is wrong and eventually how to fix that!

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/pm4py/pm4py-source/issues/75#issuecomment-483310736, or mute the thread https://github.com/notifications/unsubscribe-auth/ADUOWBB732O6FGRCO3KZPWLPQSQIBANCNFSM4HGBC7ZQ .

[lvzheqi]

And I have a technic problem by installation requirements in windows. Could you help me? I want to run the data on the computer in the uni, because it takes too long, to run the data in my own computer. So I try to setup the environment. I download visual studio and the system-operation is 64. But it always shows this error, when it installs "pyarrow". [image: image.png] [image: image.png]

吕哲奇 lvzheqi@gmail.com 于2019年4月15日周一 下午5:59写道：

For me, I think, fitness = 1 - align["cost"]) / ( len(log[index]) MODEL_MOVE_COST + best_worst_cost LOG_MOVE_COST). I'm not so sure... MODEL_MOVE_COST and LOG_MOVE_COST maybe need to extract from parameters

Javert899 notifications@github.com 于2019年4月15日周一 下午5:53写道：

Here it depends on how the fitness value is calculated :)

For this let me know if you think it is wrong and eventually how to fix that!

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/pm4py/pm4py-source/issues/75#issuecomment-483310736, or mute the thread https://github.com/notifications/unsubscribe-auth/ADUOWBB732O6FGRCO3KZPWLPQSQIBANCNFSM4HGBC7ZQ .

[Javert899]

Tomorrow I can come after lunch :)

[lvzheqi]

Super, thanks! 😄

Javert899 notifications@github.com 于2019年4月15日周一 下午6:26写道：

Tomorrow I can come after lunch :)

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/pm4py/pm4py-source/issues/75#issuecomment-483323025, or mute the thread https://github.com/notifications/unsubscribe-auth/ADUOWBBPIAQGAM3ESTCMMZTPQSUCJANCNFSM4HGBC7ZQ .

[Javert899]

Keep note that in the 'develop' branch alignments are got much faster in comparison to the 'stable' release (nothing approach-wise but a lot of tricks, they are 40% faster in each condition now, not considering variants grouping that on some logs can speed up things of 100x)

[lvzheqi]

Yep, I see. Thanks.

Javert899 notifications@github.com 于2019年4月15日周一 下午6:28写道：

Keep note that in the 'develop' branch alignments are got much faster in comparison to the 'stable' release (nothing approach-wise but a lot of tricks, they are 40% faster in each condition now, not considering variants grouping that on some logs can speed up things of 100x)

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/pm4py/pm4py-source/issues/75#issuecomment-483323775, or mute the thread https://github.com/notifications/unsubscribe-auth/ADUOWBDIUUZQIRXXQDGXLETPQSUJPANCNFSM4HGBC7ZQ .

[Javert899]

After code exchange between me and lvzheqi I would say this issue is definitely closed :)