Dev accessibilities · Pull Request #614 · ActivitySim/activitysim

ghost · 2022-10-04T15:47:43Z

Draft pull request for disaggregate accessibilities. Works with 1 and 2 zone models. Tested but some issues remain:

Optional sampling and soft-merge methods using kmeans/naive bayes require sklearn package not part of testing suite. Disabled these imports for now. However, the pre-sampling method could/should be improved to use the weighted sampling method used in activitysim.
difficulty testing multiprocess, it appears vehicle model does not pass mp test. Unsure if related or isolated issue.

…eeded to modify this file

…subprocess

…ry samples

i-am-sijia · 2022-11-10T04:42:40Z

+
+ # Assign persons to households
+ rep = (
+ pd.DataFrame(util.named_product(hhid=households[hhid], index=persons.index))


Seems like the proto population does not allow households to have different household sizes nor different household member composition. Is that intended?

i-am-sijia · 2022-11-10T04:59:54Z

+ hinccat1: [1, 2, 3, 4] # Income categories
+ hworkers: 1 # Household workers
+ veh: [0, 1, 2] # Household vehicles
+ persons: 2 # Two persons household


Can household size vary in the proto population? e.g. persons: [1, 2, 3]. And can households have different composition? e.g. type 1 household has one working adult and one school child, type 2 household has one working adult and one retiree. In applications, household size and composition could have impact on mode chocie logsum.

You're correct, this was created as specified in the 'Disaggregate Accessibilities Implementation Plan' (https://github.com/ActivitySim/activitysim/wiki/Project-Meeting-2022.03.22).
You can change the household size, but you can't have multiple household sizes. E.g., you can't have a 2 person household and a 3 person household because persons must then be mapped to the households and this greatly complicates the generation procedure.
I agree that users may want this functionality, but is a bit out of scope. However, I think a simple solution for complicated cases would be to just let users to optionally create their own person/household/tour templates tables manually in a csv that gets read in. I just added this to the latest commit.

i-am-sijia · 2022-11-10T05:12:37Z

+ grade: 0 # Not attending
+ timeFactorWork: 1 # mean
+ timeFactorNonWork: 1 # mean
+ mapped_fields:


I think users would benefit from more detailed documentation on these settings in case they want to change them. E.g. when does it require users to define mappings for all vs some attributes; what is the minimum required attributes to create/map (seems like at least the ones used in the mode choice and location choice uec specs). I could be wrong but it seems there are some underlying rules like no need to define two workers in the proto pop.

These are person attributes specified in the plan document. However, the timeFactorNonWork and timeFactorWork are not necessary to ActivitySim and have been removed. But will mention in documentation that the proto tables must include all fields necessary to run your models.

Thanks. Where can I find the plan document?

i-am-sijia · 2022-11-10T05:59:04Z

+ sample_idx = sorted(sample_idx)
+ elif method and method.lower() == "kmeans":
+ # Performs a simple k-means clustering using centroid XY coordinates
+ centroids_df = pipeline.get_table("maz_centroids")


Seems kmeans will not work with one-zone system. Can we add assert before this?

Added assertion for 2-zone systems. 1-zone could be implemented in future but may not be needed if aggressive sampling isn't typically requires for one-zone systems. This will require more work to ensure variable zoned centroid tables are loaded properly in tables/disaggregate_accessibility.py.

…mplementation, still working.

i-am-sijia · 2022-12-15T16:57:51Z

@nick-fournier-rsg Thank you for addressing my comments. One new question - Should at least one 2-zone example model be updated to include disagg accessibility? Right now disagg accessibility is only tested within the 1-zone MTC extended model.

i-am-sijia · 2022-12-30T17:48:18Z

Confirmed sandag 2-zone example ran on my end, thanks for adding it.

ghost · 2022-10-04T15:47:43Z

Draft pull request for disaggregate accessibilities. Works with 1 and 2 zone models. Tested but some issues remain:

Optional sampling and soft-merge methods using kmeans/naive bayes require sklearn package not part of testing suite. Disabled these imports for now. However, the pre-sampling method could/should be improved to use the weighted sampling method used in activitysim.
difficulty testing multiprocess, it appears vehicle model does not pass mp test. Unsure if related or isolated issue.

…eeded to modify this file

…subprocess

…ry samples

i-am-sijia · 2022-11-10T04:42:40Z

+
+ # Assign persons to households
+ rep = (
+ pd.DataFrame(util.named_product(hhid=households[hhid], index=persons.index))


Seems like the proto population does not allow households to have different household sizes nor different household member composition. Is that intended?

i-am-sijia · 2022-11-10T04:59:54Z

+ hinccat1: [1, 2, 3, 4] # Income categories
+ hworkers: 1 # Household workers
+ veh: [0, 1, 2] # Household vehicles
+ persons: 2 # Two persons household


Can household size vary in the proto population? e.g. persons: [1, 2, 3]. And can households have different composition? e.g. type 1 household has one working adult and one school child, type 2 household has one working adult and one retiree. In applications, household size and composition could have impact on mode chocie logsum.

You're correct, this was created as specified in the 'Disaggregate Accessibilities Implementation Plan' (https://github.com/ActivitySim/activitysim/wiki/Project-Meeting-2022.03.22).
You can change the household size, but you can't have multiple household sizes. E.g., you can't have a 2 person household and a 3 person household because persons must then be mapped to the households and this greatly complicates the generation procedure.
I agree that users may want this functionality, but is a bit out of scope. However, I think a simple solution for complicated cases would be to just let users to optionally create their own person/household/tour templates tables manually in a csv that gets read in. I just added this to the latest commit.

i-am-sijia · 2022-11-10T05:12:37Z

+ grade: 0 # Not attending
+ timeFactorWork: 1 # mean
+ timeFactorNonWork: 1 # mean
+ mapped_fields:


I think users would benefit from more detailed documentation on these settings in case they want to change them. E.g. when does it require users to define mappings for all vs some attributes; what is the minimum required attributes to create/map (seems like at least the ones used in the mode choice and location choice uec specs). I could be wrong but it seems there are some underlying rules like no need to define two workers in the proto pop.

These are person attributes specified in the plan document. However, the timeFactorNonWork and timeFactorWork are not necessary to ActivitySim and have been removed. But will mention in documentation that the proto tables must include all fields necessary to run your models.

Thanks. Where can I find the plan document?

i-am-sijia · 2022-11-10T05:59:04Z

+ sample_idx = sorted(sample_idx)
+ elif method and method.lower() == "kmeans":
+ # Performs a simple k-means clustering using centroid XY coordinates
+ centroids_df = pipeline.get_table("maz_centroids")


Seems kmeans will not work with one-zone system. Can we add assert before this?

Added assertion for 2-zone systems. 1-zone could be implemented in future but may not be needed if aggressive sampling isn't typically requires for one-zone systems. This will require more work to ensure variable zoned centroid tables are loaded properly in tables/disaggregate_accessibility.py.

…mplementation, still working.

i-am-sijia · 2022-12-15T16:57:51Z

@nick-fournier-rsg Thank you for addressing my comments. One new question - Should at least one 2-zone example model be updated to include disagg accessibility? Right now disagg accessibility is only tested within the 1-zone MTC extended model.

i-am-sijia · 2022-12-30T17:48:18Z

Confirmed sandag 2-zone example ran on my end, thanks for adding it.

ghost · 2022-10-04T15:47:43Z

Draft pull request for disaggregate accessibilities. Works with 1 and 2 zone models. Tested but some issues remain:

Optional sampling and soft-merge methods using kmeans/naive bayes require sklearn package not part of testing suite. Disabled these imports for now. However, the pre-sampling method could/should be improved to use the weighted sampling method used in activitysim.
difficulty testing multiprocess, it appears vehicle model does not pass mp test. Unsure if related or isolated issue.

…eeded to modify this file

…subprocess

…ry samples

i-am-sijia · 2022-11-10T04:42:40Z

+
+ # Assign persons to households
+ rep = (
+ pd.DataFrame(util.named_product(hhid=households[hhid], index=persons.index))


Seems like the proto population does not allow households to have different household sizes nor different household member composition. Is that intended?

i-am-sijia · 2022-11-10T04:59:54Z

+ hinccat1: [1, 2, 3, 4] # Income categories
+ hworkers: 1 # Household workers
+ veh: [0, 1, 2] # Household vehicles
+ persons: 2 # Two persons household


Can household size vary in the proto population? e.g. persons: [1, 2, 3]. And can households have different composition? e.g. type 1 household has one working adult and one school child, type 2 household has one working adult and one retiree. In applications, household size and composition could have impact on mode chocie logsum.

You're correct, this was created as specified in the 'Disaggregate Accessibilities Implementation Plan' (https://github.com/ActivitySim/activitysim/wiki/Project-Meeting-2022.03.22).
You can change the household size, but you can't have multiple household sizes. E.g., you can't have a 2 person household and a 3 person household because persons must then be mapped to the households and this greatly complicates the generation procedure.
I agree that users may want this functionality, but is a bit out of scope. However, I think a simple solution for complicated cases would be to just let users to optionally create their own person/household/tour templates tables manually in a csv that gets read in. I just added this to the latest commit.

i-am-sijia · 2022-11-10T05:12:37Z

+ grade: 0 # Not attending
+ timeFactorWork: 1 # mean
+ timeFactorNonWork: 1 # mean
+ mapped_fields:


I think users would benefit from more detailed documentation on these settings in case they want to change them. E.g. when does it require users to define mappings for all vs some attributes; what is the minimum required attributes to create/map (seems like at least the ones used in the mode choice and location choice uec specs). I could be wrong but it seems there are some underlying rules like no need to define two workers in the proto pop.

These are person attributes specified in the plan document. However, the timeFactorNonWork and timeFactorWork are not necessary to ActivitySim and have been removed. But will mention in documentation that the proto tables must include all fields necessary to run your models.

Thanks. Where can I find the plan document?

i-am-sijia · 2022-11-10T05:59:04Z

+ sample_idx = sorted(sample_idx)
+ elif method and method.lower() == "kmeans":
+ # Performs a simple k-means clustering using centroid XY coordinates
+ centroids_df = pipeline.get_table("maz_centroids")


Seems kmeans will not work with one-zone system. Can we add assert before this?

Added assertion for 2-zone systems. 1-zone could be implemented in future but may not be needed if aggressive sampling isn't typically requires for one-zone systems. This will require more work to ensure variable zoned centroid tables are loaded properly in tables/disaggregate_accessibility.py.

…mplementation, still working.

i-am-sijia · 2022-12-15T16:57:51Z

@nick-fournier-rsg Thank you for addressing my comments. One new question - Should at least one 2-zone example model be updated to include disagg accessibility? Right now disagg accessibility is only tested within the 1-zone MTC extended model.

i-am-sijia · 2022-12-30T17:48:18Z

Confirmed sandag 2-zone example ran on my end, thanks for adding it.

ghost · 2022-10-04T15:47:43Z

Draft pull request for disaggregate accessibilities. Works with 1 and 2 zone models. Tested but some issues remain:

Optional sampling and soft-merge methods using kmeans/naive bayes require sklearn package not part of testing suite. Disabled these imports for now. However, the pre-sampling method could/should be improved to use the weighted sampling method used in activitysim.
difficulty testing multiprocess, it appears vehicle model does not pass mp test. Unsure if related or isolated issue.

…eeded to modify this file

…subprocess

…ry samples

i-am-sijia · 2022-11-10T04:42:40Z

+
+ # Assign persons to households
+ rep = (
+ pd.DataFrame(util.named_product(hhid=households[hhid], index=persons.index))


Seems like the proto population does not allow households to have different household sizes nor different household member composition. Is that intended?

i-am-sijia · 2022-11-10T04:59:54Z

+ hinccat1: [1, 2, 3, 4] # Income categories
+ hworkers: 1 # Household workers
+ veh: [0, 1, 2] # Household vehicles
+ persons: 2 # Two persons household


Can household size vary in the proto population? e.g. persons: [1, 2, 3]. And can households have different composition? e.g. type 1 household has one working adult and one school child, type 2 household has one working adult and one retiree. In applications, household size and composition could have impact on mode chocie logsum.

You're correct, this was created as specified in the 'Disaggregate Accessibilities Implementation Plan' (https://github.com/ActivitySim/activitysim/wiki/Project-Meeting-2022.03.22).
You can change the household size, but you can't have multiple household sizes. E.g., you can't have a 2 person household and a 3 person household because persons must then be mapped to the households and this greatly complicates the generation procedure.
I agree that users may want this functionality, but is a bit out of scope. However, I think a simple solution for complicated cases would be to just let users to optionally create their own person/household/tour templates tables manually in a csv that gets read in. I just added this to the latest commit.

i-am-sijia · 2022-11-10T05:12:37Z

+ grade: 0 # Not attending
+ timeFactorWork: 1 # mean
+ timeFactorNonWork: 1 # mean
+ mapped_fields:


I think users would benefit from more detailed documentation on these settings in case they want to change them. E.g. when does it require users to define mappings for all vs some attributes; what is the minimum required attributes to create/map (seems like at least the ones used in the mode choice and location choice uec specs). I could be wrong but it seems there are some underlying rules like no need to define two workers in the proto pop.

These are person attributes specified in the plan document. However, the timeFactorNonWork and timeFactorWork are not necessary to ActivitySim and have been removed. But will mention in documentation that the proto tables must include all fields necessary to run your models.

Thanks. Where can I find the plan document?

i-am-sijia · 2022-11-10T05:59:04Z

+ sample_idx = sorted(sample_idx)
+ elif method and method.lower() == "kmeans":
+ # Performs a simple k-means clustering using centroid XY coordinates
+ centroids_df = pipeline.get_table("maz_centroids")


Seems kmeans will not work with one-zone system. Can we add assert before this?

Added assertion for 2-zone systems. 1-zone could be implemented in future but may not be needed if aggressive sampling isn't typically requires for one-zone systems. This will require more work to ensure variable zoned centroid tables are loaded properly in tables/disaggregate_accessibility.py.

…mplementation, still working.

i-am-sijia · 2022-12-15T16:57:51Z

@nick-fournier-rsg Thank you for addressing my comments. One new question - Should at least one 2-zone example model be updated to include disagg accessibility? Right now disagg accessibility is only tested within the 1-zone MTC extended model.

i-am-sijia · 2022-12-30T17:48:18Z

Confirmed sandag 2-zone example ran on my end, thanks for adding it.

ghost · 2022-10-04T15:47:43Z

Draft pull request for disaggregate accessibilities. Works with 1 and 2 zone models. Tested but some issues remain:

Optional sampling and soft-merge methods using kmeans/naive bayes require sklearn package not part of testing suite. Disabled these imports for now. However, the pre-sampling method could/should be improved to use the weighted sampling method used in activitysim.
difficulty testing multiprocess, it appears vehicle model does not pass mp test. Unsure if related or isolated issue.

…eeded to modify this file

…subprocess

…ry samples

i-am-sijia · 2022-11-10T04:42:40Z

+
+ # Assign persons to households
+ rep = (
+ pd.DataFrame(util.named_product(hhid=households[hhid], index=persons.index))


Seems like the proto population does not allow households to have different household sizes nor different household member composition. Is that intended?

i-am-sijia · 2022-11-10T04:59:54Z

+ hinccat1: [1, 2, 3, 4] # Income categories
+ hworkers: 1 # Household workers
+ veh: [0, 1, 2] # Household vehicles
+ persons: 2 # Two persons household


Can household size vary in the proto population? e.g. persons: [1, 2, 3]. And can households have different composition? e.g. type 1 household has one working adult and one school child, type 2 household has one working adult and one retiree. In applications, household size and composition could have impact on mode chocie logsum.

You're correct, this was created as specified in the 'Disaggregate Accessibilities Implementation Plan' (https://github.com/ActivitySim/activitysim/wiki/Project-Meeting-2022.03.22).
You can change the household size, but you can't have multiple household sizes. E.g., you can't have a 2 person household and a 3 person household because persons must then be mapped to the households and this greatly complicates the generation procedure.
I agree that users may want this functionality, but is a bit out of scope. However, I think a simple solution for complicated cases would be to just let users to optionally create their own person/household/tour templates tables manually in a csv that gets read in. I just added this to the latest commit.

i-am-sijia · 2022-11-10T05:12:37Z

+ grade: 0 # Not attending
+ timeFactorWork: 1 # mean
+ timeFactorNonWork: 1 # mean
+ mapped_fields:


I think users would benefit from more detailed documentation on these settings in case they want to change them. E.g. when does it require users to define mappings for all vs some attributes; what is the minimum required attributes to create/map (seems like at least the ones used in the mode choice and location choice uec specs). I could be wrong but it seems there are some underlying rules like no need to define two workers in the proto pop.

These are person attributes specified in the plan document. However, the timeFactorNonWork and timeFactorWork are not necessary to ActivitySim and have been removed. But will mention in documentation that the proto tables must include all fields necessary to run your models.

Thanks. Where can I find the plan document?

i-am-sijia · 2022-11-10T05:59:04Z

+ sample_idx = sorted(sample_idx)
+ elif method and method.lower() == "kmeans":
+ # Performs a simple k-means clustering using centroid XY coordinates
+ centroids_df = pipeline.get_table("maz_centroids")


Seems kmeans will not work with one-zone system. Can we add assert before this?

Added assertion for 2-zone systems. 1-zone could be implemented in future but may not be needed if aggressive sampling isn't typically requires for one-zone systems. This will require more work to ensure variable zoned centroid tables are loaded properly in tables/disaggregate_accessibility.py.

…mplementation, still working.

i-am-sijia · 2022-12-15T16:57:51Z

@nick-fournier-rsg Thank you for addressing my comments. One new question - Should at least one 2-zone example model be updated to include disagg accessibility? Right now disagg accessibility is only tested within the 1-zone MTC extended model.

i-am-sijia · 2022-12-30T17:48:18Z

Confirmed sandag 2-zone example ran on my end, thanks for adding it.

ghost · 2022-10-04T15:47:43Z

Draft pull request for disaggregate accessibilities. Works with 1 and 2 zone models. Tested but some issues remain:

Optional sampling and soft-merge methods using kmeans/naive bayes require sklearn package not part of testing suite. Disabled these imports for now. However, the pre-sampling method could/should be improved to use the weighted sampling method used in activitysim.
difficulty testing multiprocess, it appears vehicle model does not pass mp test. Unsure if related or isolated issue.

…eeded to modify this file

…subprocess

…ry samples

i-am-sijia · 2022-11-10T04:42:40Z

+
+ # Assign persons to households
+ rep = (
+ pd.DataFrame(util.named_product(hhid=households[hhid], index=persons.index))


Seems like the proto population does not allow households to have different household sizes nor different household member composition. Is that intended?

i-am-sijia · 2022-11-10T04:59:54Z

+ hinccat1: [1, 2, 3, 4] # Income categories
+ hworkers: 1 # Household workers
+ veh: [0, 1, 2] # Household vehicles
+ persons: 2 # Two persons household


Can household size vary in the proto population? e.g. persons: [1, 2, 3]. And can households have different composition? e.g. type 1 household has one working adult and one school child, type 2 household has one working adult and one retiree. In applications, household size and composition could have impact on mode chocie logsum.

You're correct, this was created as specified in the 'Disaggregate Accessibilities Implementation Plan' (https://github.com/ActivitySim/activitysim/wiki/Project-Meeting-2022.03.22).
You can change the household size, but you can't have multiple household sizes. E.g., you can't have a 2 person household and a 3 person household because persons must then be mapped to the households and this greatly complicates the generation procedure.
I agree that users may want this functionality, but is a bit out of scope. However, I think a simple solution for complicated cases would be to just let users to optionally create their own person/household/tour templates tables manually in a csv that gets read in. I just added this to the latest commit.

i-am-sijia · 2022-11-10T05:12:37Z

+ grade: 0 # Not attending
+ timeFactorWork: 1 # mean
+ timeFactorNonWork: 1 # mean
+ mapped_fields:


I think users would benefit from more detailed documentation on these settings in case they want to change them. E.g. when does it require users to define mappings for all vs some attributes; what is the minimum required attributes to create/map (seems like at least the ones used in the mode choice and location choice uec specs). I could be wrong but it seems there are some underlying rules like no need to define two workers in the proto pop.

These are person attributes specified in the plan document. However, the timeFactorNonWork and timeFactorWork are not necessary to ActivitySim and have been removed. But will mention in documentation that the proto tables must include all fields necessary to run your models.

Thanks. Where can I find the plan document?

i-am-sijia · 2022-11-10T05:59:04Z

+ sample_idx = sorted(sample_idx)
+ elif method and method.lower() == "kmeans":
+ # Performs a simple k-means clustering using centroid XY coordinates
+ centroids_df = pipeline.get_table("maz_centroids")


Seems kmeans will not work with one-zone system. Can we add assert before this?

Added assertion for 2-zone systems. 1-zone could be implemented in future but may not be needed if aggressive sampling isn't typically requires for one-zone systems. This will require more work to ensure variable zoned centroid tables are loaded properly in tables/disaggregate_accessibility.py.

…mplementation, still working.

i-am-sijia · 2022-12-15T16:57:51Z

@nick-fournier-rsg Thank you for addressing my comments. One new question - Should at least one 2-zone example model be updated to include disagg accessibility? Right now disagg accessibility is only tested within the 1-zone MTC extended model.

i-am-sijia · 2022-12-30T17:48:18Z

Confirmed sandag 2-zone example ran on my end, thanks for adding it.

ghost · 2022-10-04T15:47:43Z

Draft pull request for disaggregate accessibilities. Works with 1 and 2 zone models. Tested but some issues remain:

Optional sampling and soft-merge methods using kmeans/naive bayes require sklearn package not part of testing suite. Disabled these imports for now. However, the pre-sampling method could/should be improved to use the weighted sampling method used in activitysim.
difficulty testing multiprocess, it appears vehicle model does not pass mp test. Unsure if related or isolated issue.

…eeded to modify this file

…subprocess

…ry samples

i-am-sijia · 2022-11-10T04:42:40Z

+
+ # Assign persons to households
+ rep = (
+ pd.DataFrame(util.named_product(hhid=households[hhid], index=persons.index))


Seems like the proto population does not allow households to have different household sizes nor different household member composition. Is that intended?

i-am-sijia · 2022-11-10T04:59:54Z

+ hinccat1: [1, 2, 3, 4] # Income categories
+ hworkers: 1 # Household workers
+ veh: [0, 1, 2] # Household vehicles
+ persons: 2 # Two persons household


Can household size vary in the proto population? e.g. persons: [1, 2, 3]. And can households have different composition? e.g. type 1 household has one working adult and one school child, type 2 household has one working adult and one retiree. In applications, household size and composition could have impact on mode chocie logsum.

You're correct, this was created as specified in the 'Disaggregate Accessibilities Implementation Plan' (https://github.com/ActivitySim/activitysim/wiki/Project-Meeting-2022.03.22).
You can change the household size, but you can't have multiple household sizes. E.g., you can't have a 2 person household and a 3 person household because persons must then be mapped to the households and this greatly complicates the generation procedure.
I agree that users may want this functionality, but is a bit out of scope. However, I think a simple solution for complicated cases would be to just let users to optionally create their own person/household/tour templates tables manually in a csv that gets read in. I just added this to the latest commit.

i-am-sijia · 2022-11-10T05:12:37Z

+ grade: 0 # Not attending
+ timeFactorWork: 1 # mean
+ timeFactorNonWork: 1 # mean
+ mapped_fields:


I think users would benefit from more detailed documentation on these settings in case they want to change them. E.g. when does it require users to define mappings for all vs some attributes; what is the minimum required attributes to create/map (seems like at least the ones used in the mode choice and location choice uec specs). I could be wrong but it seems there are some underlying rules like no need to define two workers in the proto pop.

These are person attributes specified in the plan document. However, the timeFactorNonWork and timeFactorWork are not necessary to ActivitySim and have been removed. But will mention in documentation that the proto tables must include all fields necessary to run your models.

Thanks. Where can I find the plan document?

i-am-sijia · 2022-11-10T05:59:04Z

+ sample_idx = sorted(sample_idx)
+ elif method and method.lower() == "kmeans":
+ # Performs a simple k-means clustering using centroid XY coordinates
+ centroids_df = pipeline.get_table("maz_centroids")


Seems kmeans will not work with one-zone system. Can we add assert before this?

Added assertion for 2-zone systems. 1-zone could be implemented in future but may not be needed if aggressive sampling isn't typically requires for one-zone systems. This will require more work to ensure variable zoned centroid tables are loaded properly in tables/disaggregate_accessibility.py.

…mplementation, still working.

i-am-sijia · 2022-12-15T16:57:51Z

@nick-fournier-rsg Thank you for addressing my comments. One new question - Should at least one 2-zone example model be updated to include disagg accessibility? Right now disagg accessibility is only tested within the 1-zone MTC extended model.

i-am-sijia · 2022-12-30T17:48:18Z

Confirmed sandag 2-zone example ran on my end, thanks for adding it.

ghost · 2022-10-04T15:47:43Z

Draft pull request for disaggregate accessibilities. Works with 1 and 2 zone models. Tested but some issues remain:

Optional sampling and soft-merge methods using kmeans/naive bayes require sklearn package not part of testing suite. Disabled these imports for now. However, the pre-sampling method could/should be improved to use the weighted sampling method used in activitysim.
difficulty testing multiprocess, it appears vehicle model does not pass mp test. Unsure if related or isolated issue.

…eeded to modify this file

…subprocess

…ry samples

i-am-sijia · 2022-11-10T04:42:40Z

+
+ # Assign persons to households
+ rep = (
+ pd.DataFrame(util.named_product(hhid=households[hhid], index=persons.index))


Seems like the proto population does not allow households to have different household sizes nor different household member composition. Is that intended?

i-am-sijia · 2022-11-10T04:59:54Z

+ hinccat1: [1, 2, 3, 4] # Income categories
+ hworkers: 1 # Household workers
+ veh: [0, 1, 2] # Household vehicles
+ persons: 2 # Two persons household


Can household size vary in the proto population? e.g. persons: [1, 2, 3]. And can households have different composition? e.g. type 1 household has one working adult and one school child, type 2 household has one working adult and one retiree. In applications, household size and composition could have impact on mode chocie logsum.

You're correct, this was created as specified in the 'Disaggregate Accessibilities Implementation Plan' (https://github.com/ActivitySim/activitysim/wiki/Project-Meeting-2022.03.22).
You can change the household size, but you can't have multiple household sizes. E.g., you can't have a 2 person household and a 3 person household because persons must then be mapped to the households and this greatly complicates the generation procedure.
I agree that users may want this functionality, but is a bit out of scope. However, I think a simple solution for complicated cases would be to just let users to optionally create their own person/household/tour templates tables manually in a csv that gets read in. I just added this to the latest commit.

i-am-sijia · 2022-11-10T05:12:37Z

+ grade: 0 # Not attending
+ timeFactorWork: 1 # mean
+ timeFactorNonWork: 1 # mean
+ mapped_fields:


I think users would benefit from more detailed documentation on these settings in case they want to change them. E.g. when does it require users to define mappings for all vs some attributes; what is the minimum required attributes to create/map (seems like at least the ones used in the mode choice and location choice uec specs). I could be wrong but it seems there are some underlying rules like no need to define two workers in the proto pop.

These are person attributes specified in the plan document. However, the timeFactorNonWork and timeFactorWork are not necessary to ActivitySim and have been removed. But will mention in documentation that the proto tables must include all fields necessary to run your models.

Thanks. Where can I find the plan document?

i-am-sijia · 2022-11-10T05:59:04Z

+ sample_idx = sorted(sample_idx)
+ elif method and method.lower() == "kmeans":
+ # Performs a simple k-means clustering using centroid XY coordinates
+ centroids_df = pipeline.get_table("maz_centroids")


Seems kmeans will not work with one-zone system. Can we add assert before this?

Added assertion for 2-zone systems. 1-zone could be implemented in future but may not be needed if aggressive sampling isn't typically requires for one-zone systems. This will require more work to ensure variable zoned centroid tables are loaded properly in tables/disaggregate_accessibility.py.

…mplementation, still working.

i-am-sijia · 2022-12-15T16:57:51Z

@nick-fournier-rsg Thank you for addressing my comments. One new question - Should at least one 2-zone example model be updated to include disagg accessibility? Right now disagg accessibility is only tested within the 1-zone MTC extended model.

i-am-sijia · 2022-12-30T17:48:18Z

Confirmed sandag 2-zone example ran on my end, thanks for adding it.

nick-fournier added 30 commits June 29, 2022 12:05

initial commit with basic working proto pop gen
b3df946

fixed bad dependency range
6441c9a

fixed vary_on method to vary all with mapped fields after
1b8511f

added check if already dataframe
96fca8e

added dataframe check if the table passed is already a dataframe
e92ed4a

working proto pop gen. still debugging mandatory work/school model
68ff25e

check all table is df
5189fc8

workplace runs, next testing school
a8c2376

began setting up to get all logsums, not just chosen ones
c82504f

restructured to run as either one-off or model step
b9fa41b

revert to last merge. Removed dataframe check in iterate, no longer n…
7838d2f
…eeded to modify this file

extracted logsums, need to inject into pipeline or save output?
286b5c9

working standalone or step model
c835590

extractable accessibilities for fixed and nonmandatory
a73e6a4

extractable accessibilities for fixed and nonmandatory
194d327

cleanup and add documentation
a237cfd

resolve minor conflicts in model __init__ and conda env yml
d5a7bdd

resolved duplicates of old 'example' folders
2e993d9

model uses write_tables functionality
3bb82b1

disaggregate accessibility runs as model step or standalone. Runs as …
6791a43
…subprocess

working model test. table initialization not yet working
3022f67

added logsum short circuit to get logsums and avoid drawing unneccesa…
256b778
…ry samples

working model plus major cleanup
ee3a1f1

working model plus major cleanup
69efa7b

override sample size
c6f26b5

fixed trace output files
a26902c

skip_choice default to False
c877a46

fixed empty logsums bug
765fd74

cleanup redundent parameters
5b27210

fixed list index instead of set
a2137c3

fixed typo
5960f40

dhensle requested a review from i-am-sijia October 20, 2022 16:31

nick-fournierand others added 7 commits October 25, 2022 11:08

update sampling script
6584f7b

'blacken'
dada5cd

fixed replacement sample bug
1096cfb

updated documentation
0d0a7c3

more flexible scenario testing
17c6b21

blacken
7dde020

added test runs to __main__ in test_mtc_extended
73264ab

dhensle marked this pull request as ready for review October 31, 2022 17:41

additional logging, random seed option, landuse sample weight column
f5f4a2e

i-am-sijia reviewed Nov 10, 2022
View reviewed changes

Comment threadactivitysim/abm/tables/disaggregate_accessibility.py

i-am-sijia reviewed Nov 10, 2022
View reviewed changes

i-am-sijia reviewed Nov 18, 2022
View reviewed changes

Comment threadactivitysim/examples/prototype_mtc_extended/configs/disaggregate_accessibility.yaml

nick-fournier added 6 commits December 7, 2022 12:01

revisions per Sijia's comments, except for the proto-table-template i…
bb813c0
…mplementation, still working.

added template pop option
f738aa3

blacken
ce78cfb

set back to create_tables to pass test against old tables
f02c466

updated proto-pop test data
57e726a

resolved bad test data issue
f0b6524

jpn-- added this to the Phase 7 milestone Dec 13, 2022

dhensleand others added 2 commits December 14, 2022 16:52

optimize nearest zone determination
6a845e4

Merge branch 'develop' into dev_accessibilities
902cb81

adding sandag 2-zone test
d132ac2

Merge branch 'develop' into dev_accessibilities
9db0029

jpn-- mentioned this pull request Jan 27, 2023
Composite - Disaggregate Accessibility #635
Merged

jpn-- merged commit f800b95 into ActivitySim:developJan 28, 2023

nick-fournier added 30 commits June 29, 2022 12:05

initial commit with basic working proto pop gen
b3df946

fixed bad dependency range
6441c9a

fixed vary_on method to vary all with mapped fields after
1b8511f

added check if already dataframe
96fca8e

added dataframe check if the table passed is already a dataframe
e92ed4a

working proto pop gen. still debugging mandatory work/school model
68ff25e

check all table is df
5189fc8

workplace runs, next testing school
a8c2376

began setting up to get all logsums, not just chosen ones
c82504f

restructured to run as either one-off or model step
b9fa41b

revert to last merge. Removed dataframe check in iterate, no longer n…
7838d2f
…eeded to modify this file

extracted logsums, need to inject into pipeline or save output?
286b5c9

working standalone or step model
c835590

extractable accessibilities for fixed and nonmandatory
a73e6a4

extractable accessibilities for fixed and nonmandatory
194d327

cleanup and add documentation
a237cfd

resolve minor conflicts in model __init__ and conda env yml
d5a7bdd

resolved duplicates of old 'example' folders
2e993d9

model uses write_tables functionality
3bb82b1

disaggregate accessibility runs as model step or standalone. Runs as …
6791a43
…subprocess

working model test. table initialization not yet working
3022f67

added logsum short circuit to get logsums and avoid drawing unneccesa…
256b778
…ry samples

working model plus major cleanup
ee3a1f1

working model plus major cleanup
69efa7b

override sample size
c6f26b5

fixed trace output files
a26902c

skip_choice default to False
c877a46

fixed empty logsums bug
765fd74

cleanup redundent parameters
5b27210

fixed list index instead of set
a2137c3

fixed typo
5960f40

dhensle requested a review from i-am-sijia October 20, 2022 16:31

nick-fournierand others added 7 commits October 25, 2022 11:08

update sampling script
6584f7b

'blacken'
dada5cd

fixed replacement sample bug
1096cfb

updated documentation
0d0a7c3

more flexible scenario testing
17c6b21

blacken
7dde020

added test runs to __main__ in test_mtc_extended
73264ab

dhensle marked this pull request as ready for review October 31, 2022 17:41

additional logging, random seed option, landuse sample weight column
f5f4a2e

i-am-sijia reviewed Nov 10, 2022
View reviewed changes

Comment threadactivitysim/abm/tables/disaggregate_accessibility.py

i-am-sijia reviewed Nov 10, 2022
View reviewed changes

i-am-sijia reviewed Nov 18, 2022
View reviewed changes

Comment threadactivitysim/examples/prototype_mtc_extended/configs/disaggregate_accessibility.yaml

nick-fournier added 6 commits December 7, 2022 12:01

revisions per Sijia's comments, except for the proto-table-template i…
bb813c0
…mplementation, still working.

added template pop option
f738aa3

blacken
ce78cfb

set back to create_tables to pass test against old tables
f02c466

updated proto-pop test data
57e726a

resolved bad test data issue
f0b6524

jpn-- added this to the Phase 7 milestone Dec 13, 2022

dhensleand others added 2 commits December 14, 2022 16:52

optimize nearest zone determination
6a845e4

Merge branch 'develop' into dev_accessibilities
902cb81

adding sandag 2-zone test
d132ac2

Merge branch 'develop' into dev_accessibilities
9db0029

jpn-- mentioned this pull request Jan 27, 2023
Composite - Disaggregate Accessibility #635
Merged

jpn-- merged commit f800b95 into ActivitySim:developJan 28, 2023

nick-fournier added 30 commits June 29, 2022 12:05

initial commit with basic working proto pop gen
b3df946

fixed bad dependency range
6441c9a

fixed vary_on method to vary all with mapped fields after
1b8511f

added check if already dataframe
96fca8e

added dataframe check if the table passed is already a dataframe
e92ed4a

working proto pop gen. still debugging mandatory work/school model
68ff25e

check all table is df
5189fc8

workplace runs, next testing school
a8c2376

began setting up to get all logsums, not just chosen ones
c82504f

restructured to run as either one-off or model step
b9fa41b

revert to last merge. Removed dataframe check in iterate, no longer n…
7838d2f
…eeded to modify this file

extracted logsums, need to inject into pipeline or save output?
286b5c9

working standalone or step model
c835590

extractable accessibilities for fixed and nonmandatory
a73e6a4

extractable accessibilities for fixed and nonmandatory
194d327

cleanup and add documentation
a237cfd

resolve minor conflicts in model __init__ and conda env yml
d5a7bdd

resolved duplicates of old 'example' folders
2e993d9

model uses write_tables functionality
3bb82b1

disaggregate accessibility runs as model step or standalone. Runs as …
6791a43
…subprocess

working model test. table initialization not yet working
3022f67

added logsum short circuit to get logsums and avoid drawing unneccesa…
256b778
…ry samples

working model plus major cleanup
ee3a1f1

working model plus major cleanup
69efa7b

override sample size
c6f26b5

fixed trace output files
a26902c

skip_choice default to False
c877a46

fixed empty logsums bug
765fd74

cleanup redundent parameters
5b27210

fixed list index instead of set
a2137c3

fixed typo
5960f40

dhensle requested a review from i-am-sijia October 20, 2022 16:31

nick-fournierand others added 7 commits October 25, 2022 11:08

update sampling script
6584f7b

'blacken'
dada5cd

fixed replacement sample bug
1096cfb

updated documentation
0d0a7c3

more flexible scenario testing
17c6b21

blacken
7dde020

added test runs to __main__ in test_mtc_extended
73264ab

dhensle marked this pull request as ready for review October 31, 2022 17:41

additional logging, random seed option, landuse sample weight column
f5f4a2e

i-am-sijia reviewed Nov 10, 2022
View reviewed changes

Comment threadactivitysim/abm/tables/disaggregate_accessibility.py

i-am-sijia reviewed Nov 10, 2022
View reviewed changes

i-am-sijia reviewed Nov 18, 2022
View reviewed changes

Comment threadactivitysim/examples/prototype_mtc_extended/configs/disaggregate_accessibility.yaml

nick-fournier added 6 commits December 7, 2022 12:01

revisions per Sijia's comments, except for the proto-table-template i…
bb813c0
…mplementation, still working.

added template pop option
f738aa3

blacken
ce78cfb

set back to create_tables to pass test against old tables
f02c466

updated proto-pop test data
57e726a

resolved bad test data issue
f0b6524

jpn-- added this to the Phase 7 milestone Dec 13, 2022

dhensleand others added 2 commits December 14, 2022 16:52

optimize nearest zone determination
6a845e4

Merge branch 'develop' into dev_accessibilities
902cb81

adding sandag 2-zone test
d132ac2

Merge branch 'develop' into dev_accessibilities
9db0029

jpn-- mentioned this pull request Jan 27, 2023
Composite - Disaggregate Accessibility #635
Merged

jpn-- merged commit f800b95 into ActivitySim:developJan 28, 2023

nick-fournier added 30 commits June 29, 2022 12:05

initial commit with basic working proto pop gen
b3df946

fixed bad dependency range
6441c9a

fixed vary_on method to vary all with mapped fields after
1b8511f

added check if already dataframe
96fca8e

added dataframe check if the table passed is already a dataframe
e92ed4a

working proto pop gen. still debugging mandatory work/school model
68ff25e

check all table is df
5189fc8

workplace runs, next testing school
a8c2376

began setting up to get all logsums, not just chosen ones
c82504f

restructured to run as either one-off or model step
b9fa41b

revert to last merge. Removed dataframe check in iterate, no longer n…
7838d2f
…eeded to modify this file

extracted logsums, need to inject into pipeline or save output?
286b5c9

working standalone or step model
c835590

extractable accessibilities for fixed and nonmandatory
a73e6a4

extractable accessibilities for fixed and nonmandatory
194d327

cleanup and add documentation
a237cfd

resolve minor conflicts in model __init__ and conda env yml
d5a7bdd

resolved duplicates of old 'example' folders
2e993d9

model uses write_tables functionality
3bb82b1

disaggregate accessibility runs as model step or standalone. Runs as …
6791a43
…subprocess

working model test. table initialization not yet working
3022f67

added logsum short circuit to get logsums and avoid drawing unneccesa…
256b778
…ry samples

working model plus major cleanup
ee3a1f1

working model plus major cleanup
69efa7b

override sample size
c6f26b5

fixed trace output files
a26902c

skip_choice default to False
c877a46

fixed empty logsums bug
765fd74

cleanup redundent parameters
5b27210

fixed list index instead of set
a2137c3

fixed typo
5960f40

dhensle requested a review from i-am-sijia October 20, 2022 16:31

nick-fournierand others added 7 commits October 25, 2022 11:08

update sampling script
6584f7b

'blacken'
dada5cd

fixed replacement sample bug
1096cfb

updated documentation
0d0a7c3

more flexible scenario testing
17c6b21

blacken
7dde020

added test runs to __main__ in test_mtc_extended
73264ab

dhensle marked this pull request as ready for review October 31, 2022 17:41

additional logging, random seed option, landuse sample weight column
f5f4a2e

i-am-sijia reviewed Nov 10, 2022
View reviewed changes

Comment threadactivitysim/abm/tables/disaggregate_accessibility.py

i-am-sijia reviewed Nov 10, 2022
View reviewed changes

i-am-sijia reviewed Nov 18, 2022
View reviewed changes

Comment threadactivitysim/examples/prototype_mtc_extended/configs/disaggregate_accessibility.yaml

nick-fournier added 6 commits December 7, 2022 12:01

revisions per Sijia's comments, except for the proto-table-template i…
bb813c0
…mplementation, still working.

added template pop option
f738aa3

blacken
ce78cfb

set back to create_tables to pass test against old tables
f02c466

updated proto-pop test data
57e726a

resolved bad test data issue
f0b6524

jpn-- added this to the Phase 7 milestone Dec 13, 2022

dhensleand others added 2 commits December 14, 2022 16:52

optimize nearest zone determination
6a845e4

Merge branch 'develop' into dev_accessibilities
902cb81

adding sandag 2-zone test
d132ac2

Merge branch 'develop' into dev_accessibilities
9db0029

jpn-- mentioned this pull request Jan 27, 2023
Composite - Disaggregate Accessibility #635
Merged

jpn-- merged commit f800b95 into ActivitySim:developJan 28, 2023

nick-fournier added 30 commits June 29, 2022 12:05

initial commit with basic working proto pop gen
b3df946

fixed bad dependency range
6441c9a

fixed vary_on method to vary all with mapped fields after
1b8511f

added check if already dataframe
96fca8e

added dataframe check if the table passed is already a dataframe
e92ed4a

working proto pop gen. still debugging mandatory work/school model
68ff25e

check all table is df
5189fc8

workplace runs, next testing school
a8c2376

began setting up to get all logsums, not just chosen ones
c82504f

restructured to run as either one-off or model step
b9fa41b

revert to last merge. Removed dataframe check in iterate, no longer n…
7838d2f
…eeded to modify this file

extracted logsums, need to inject into pipeline or save output?
286b5c9

working standalone or step model
c835590

extractable accessibilities for fixed and nonmandatory
a73e6a4

extractable accessibilities for fixed and nonmandatory
194d327

cleanup and add documentation
a237cfd

resolve minor conflicts in model __init__ and conda env yml
d5a7bdd

resolved duplicates of old 'example' folders
2e993d9

model uses write_tables functionality
3bb82b1

disaggregate accessibility runs as model step or standalone. Runs as …
6791a43
…subprocess

working model test. table initialization not yet working
3022f67

added logsum short circuit to get logsums and avoid drawing unneccesa…
256b778
…ry samples

working model plus major cleanup
ee3a1f1

working model plus major cleanup
69efa7b

override sample size
c6f26b5

fixed trace output files
a26902c

skip_choice default to False
c877a46

fixed empty logsums bug
765fd74

cleanup redundent parameters
5b27210

fixed list index instead of set
a2137c3

fixed typo
5960f40

dhensle requested a review from i-am-sijia October 20, 2022 16:31

nick-fournierand others added 7 commits October 25, 2022 11:08

update sampling script
6584f7b

'blacken'
dada5cd

fixed replacement sample bug
1096cfb

updated documentation
0d0a7c3

more flexible scenario testing
17c6b21

blacken
7dde020

added test runs to __main__ in test_mtc_extended
73264ab

dhensle marked this pull request as ready for review October 31, 2022 17:41

additional logging, random seed option, landuse sample weight column
f5f4a2e

i-am-sijia reviewed Nov 10, 2022
View reviewed changes

Comment threadactivitysim/abm/tables/disaggregate_accessibility.py

i-am-sijia reviewed Nov 10, 2022
View reviewed changes

i-am-sijia reviewed Nov 18, 2022
View reviewed changes

Comment threadactivitysim/examples/prototype_mtc_extended/configs/disaggregate_accessibility.yaml

nick-fournier added 6 commits December 7, 2022 12:01

revisions per Sijia's comments, except for the proto-table-template i…
bb813c0
…mplementation, still working.

added template pop option
f738aa3

blacken
ce78cfb

set back to create_tables to pass test against old tables
f02c466

updated proto-pop test data
57e726a

resolved bad test data issue
f0b6524

jpn-- added this to the Phase 7 milestone Dec 13, 2022

dhensleand others added 2 commits December 14, 2022 16:52

optimize nearest zone determination
6a845e4

Merge branch 'develop' into dev_accessibilities
902cb81

adding sandag 2-zone test
d132ac2

Merge branch 'develop' into dev_accessibilities
9db0029

jpn-- mentioned this pull request Jan 27, 2023
Composite - Disaggregate Accessibility #635
Merged

jpn-- merged commit f800b95 into ActivitySim:developJan 28, 2023

nick-fournier added 30 commits June 29, 2022 12:05

initial commit with basic working proto pop gen
b3df946

fixed bad dependency range
6441c9a

fixed vary_on method to vary all with mapped fields after
1b8511f

added check if already dataframe
96fca8e

added dataframe check if the table passed is already a dataframe
e92ed4a

working proto pop gen. still debugging mandatory work/school model
68ff25e

check all table is df
5189fc8

workplace runs, next testing school
a8c2376

began setting up to get all logsums, not just chosen ones
c82504f

restructured to run as either one-off or model step
b9fa41b

revert to last merge. Removed dataframe check in iterate, no longer n…
7838d2f
…eeded to modify this file

extracted logsums, need to inject into pipeline or save output?
286b5c9

working standalone or step model
c835590

extractable accessibilities for fixed and nonmandatory
a73e6a4

extractable accessibilities for fixed and nonmandatory
194d327

cleanup and add documentation
a237cfd

resolve minor conflicts in model __init__ and conda env yml
d5a7bdd

resolved duplicates of old 'example' folders
2e993d9

model uses write_tables functionality
3bb82b1

disaggregate accessibility runs as model step or standalone. Runs as …
6791a43
…subprocess

working model test. table initialization not yet working
3022f67

added logsum short circuit to get logsums and avoid drawing unneccesa…
256b778
…ry samples

working model plus major cleanup
ee3a1f1

working model plus major cleanup
69efa7b

override sample size
c6f26b5

fixed trace output files
a26902c

skip_choice default to False
c877a46

fixed empty logsums bug
765fd74

cleanup redundent parameters
5b27210

fixed list index instead of set
a2137c3

fixed typo
5960f40

dhensle requested a review from i-am-sijia October 20, 2022 16:31

nick-fournierand others added 7 commits October 25, 2022 11:08

update sampling script
6584f7b

'blacken'
dada5cd

fixed replacement sample bug
1096cfb

updated documentation
0d0a7c3

more flexible scenario testing
17c6b21

blacken
7dde020

added test runs to __main__ in test_mtc_extended
73264ab

dhensle marked this pull request as ready for review October 31, 2022 17:41

additional logging, random seed option, landuse sample weight column
f5f4a2e

i-am-sijia reviewed Nov 10, 2022
View reviewed changes

Comment threadactivitysim/abm/tables/disaggregate_accessibility.py

i-am-sijia reviewed Nov 10, 2022
View reviewed changes

i-am-sijia reviewed Nov 18, 2022
View reviewed changes

Comment threadactivitysim/examples/prototype_mtc_extended/configs/disaggregate_accessibility.yaml

nick-fournier added 6 commits December 7, 2022 12:01

revisions per Sijia's comments, except for the proto-table-template i…
bb813c0
…mplementation, still working.

added template pop option
f738aa3

blacken
ce78cfb

set back to create_tables to pass test against old tables
f02c466

updated proto-pop test data
57e726a

resolved bad test data issue
f0b6524

jpn-- added this to the Phase 7 milestone Dec 13, 2022

dhensleand others added 2 commits December 14, 2022 16:52

optimize nearest zone determination
6a845e4

Merge branch 'develop' into dev_accessibilities
902cb81

adding sandag 2-zone test
d132ac2

Merge branch 'develop' into dev_accessibilities
9db0029

jpn-- mentioned this pull request Jan 27, 2023
Composite - Disaggregate Accessibility #635
Merged

jpn-- merged commit f800b95 into ActivitySim:developJan 28, 2023

nick-fournier added 30 commits June 29, 2022 12:05

initial commit with basic working proto pop gen
b3df946

fixed bad dependency range
6441c9a

fixed vary_on method to vary all with mapped fields after
1b8511f

added check if already dataframe
96fca8e

added dataframe check if the table passed is already a dataframe
e92ed4a

working proto pop gen. still debugging mandatory work/school model
68ff25e

check all table is df
5189fc8

workplace runs, next testing school
a8c2376

began setting up to get all logsums, not just chosen ones
c82504f

restructured to run as either one-off or model step
b9fa41b

revert to last merge. Removed dataframe check in iterate, no longer n…
7838d2f
…eeded to modify this file

extracted logsums, need to inject into pipeline or save output?
286b5c9

working standalone or step model
c835590

extractable accessibilities for fixed and nonmandatory
a73e6a4

extractable accessibilities for fixed and nonmandatory
194d327

cleanup and add documentation
a237cfd

resolve minor conflicts in model __init__ and conda env yml
d5a7bdd

resolved duplicates of old 'example' folders
2e993d9

model uses write_tables functionality
3bb82b1

disaggregate accessibility runs as model step or standalone. Runs as …
6791a43
…subprocess

working model test. table initialization not yet working
3022f67

added logsum short circuit to get logsums and avoid drawing unneccesa…
256b778
…ry samples

working model plus major cleanup
ee3a1f1

working model plus major cleanup
69efa7b

override sample size
c6f26b5

fixed trace output files
a26902c

skip_choice default to False
c877a46

fixed empty logsums bug
765fd74

cleanup redundent parameters
5b27210

fixed list index instead of set
a2137c3

fixed typo
5960f40

dhensle requested a review from i-am-sijia October 20, 2022 16:31

nick-fournierand others added 7 commits October 25, 2022 11:08

update sampling script
6584f7b

'blacken'
dada5cd

fixed replacement sample bug
1096cfb

updated documentation
0d0a7c3

more flexible scenario testing
17c6b21

blacken
7dde020

added test runs to __main__ in test_mtc_extended
73264ab

dhensle marked this pull request as ready for review October 31, 2022 17:41

additional logging, random seed option, landuse sample weight column
f5f4a2e

i-am-sijia reviewed Nov 10, 2022
View reviewed changes

Comment threadactivitysim/abm/tables/disaggregate_accessibility.py

i-am-sijia reviewed Nov 10, 2022
View reviewed changes

i-am-sijia reviewed Nov 18, 2022
View reviewed changes

Comment threadactivitysim/examples/prototype_mtc_extended/configs/disaggregate_accessibility.yaml

nick-fournier added 6 commits December 7, 2022 12:01

revisions per Sijia's comments, except for the proto-table-template i…
bb813c0
…mplementation, still working.

added template pop option
f738aa3

blacken
ce78cfb

set back to create_tables to pass test against old tables
f02c466

updated proto-pop test data
57e726a

resolved bad test data issue
f0b6524

jpn-- added this to the Phase 7 milestone Dec 13, 2022

dhensleand others added 2 commits December 14, 2022 16:52

optimize nearest zone determination
6a845e4

Merge branch 'develop' into dev_accessibilities
902cb81

adding sandag 2-zone test
d132ac2

Merge branch 'develop' into dev_accessibilities
9db0029

jpn-- mentioned this pull request Jan 27, 2023
Composite - Disaggregate Accessibility #635
Merged

jpn-- merged commit f800b95 into ActivitySim:developJan 28, 2023

nick-fournier added 30 commits June 29, 2022 12:05

initial commit with basic working proto pop gen
b3df946

fixed bad dependency range
6441c9a

fixed vary_on method to vary all with mapped fields after
1b8511f

added check if already dataframe
96fca8e

added dataframe check if the table passed is already a dataframe
e92ed4a

working proto pop gen. still debugging mandatory work/school model
68ff25e

check all table is df
5189fc8

workplace runs, next testing school
a8c2376

began setting up to get all logsums, not just chosen ones
c82504f

restructured to run as either one-off or model step
b9fa41b

revert to last merge. Removed dataframe check in iterate, no longer n…
7838d2f
…eeded to modify this file

extracted logsums, need to inject into pipeline or save output?
286b5c9

working standalone or step model
c835590

extractable accessibilities for fixed and nonmandatory
a73e6a4

extractable accessibilities for fixed and nonmandatory
194d327

cleanup and add documentation
a237cfd

resolve minor conflicts in model __init__ and conda env yml
d5a7bdd

resolved duplicates of old 'example' folders
2e993d9

model uses write_tables functionality
3bb82b1

disaggregate accessibility runs as model step or standalone. Runs as …
6791a43
…subprocess

working model test. table initialization not yet working
3022f67

added logsum short circuit to get logsums and avoid drawing unneccesa…
256b778
…ry samples

working model plus major cleanup
ee3a1f1

working model plus major cleanup
69efa7b

override sample size
c6f26b5

fixed trace output files
a26902c

skip_choice default to False
c877a46

fixed empty logsums bug
765fd74

cleanup redundent parameters
5b27210

fixed list index instead of set
a2137c3

fixed typo
5960f40

dhensle requested a review from i-am-sijia October 20, 2022 16:31

nick-fournierand others added 7 commits October 25, 2022 11:08

update sampling script
6584f7b

'blacken'
dada5cd

fixed replacement sample bug
1096cfb

updated documentation
0d0a7c3

more flexible scenario testing
17c6b21

blacken
7dde020

added test runs to __main__ in test_mtc_extended
73264ab

dhensle marked this pull request as ready for review October 31, 2022 17:41

additional logging, random seed option, landuse sample weight column
f5f4a2e

i-am-sijia reviewed Nov 10, 2022
View reviewed changes

Comment threadactivitysim/abm/tables/disaggregate_accessibility.py

i-am-sijia reviewed Nov 10, 2022
View reviewed changes

i-am-sijia reviewed Nov 18, 2022
View reviewed changes

Comment threadactivitysim/examples/prototype_mtc_extended/configs/disaggregate_accessibility.yaml

nick-fournier added 6 commits December 7, 2022 12:01

revisions per Sijia's comments, except for the proto-table-template i…
bb813c0
…mplementation, still working.

added template pop option
f738aa3

blacken
ce78cfb

set back to create_tables to pass test against old tables
f02c466

updated proto-pop test data
57e726a

resolved bad test data issue
f0b6524

jpn-- added this to the Phase 7 milestone Dec 13, 2022

dhensleand others added 2 commits December 14, 2022 16:52

optimize nearest zone determination
6a845e4

Merge branch 'develop' into dev_accessibilities
902cb81

adding sandag 2-zone test
d132ac2

Merge branch 'develop' into dev_accessibilities
9db0029

jpn-- mentioned this pull request Jan 27, 2023
Composite - Disaggregate Accessibility #635
Merged

jpn-- merged commit f800b95 into ActivitySim:developJan 28, 2023

Uh oh!

Conversation

ghost commented Oct 4, 2022• edited by ghost Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ghostDec 7, 2022• edited by ghost Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ghostDec 7, 2022• edited by ghost Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

i-am-sijia commented Dec 15, 2022

Uh oh!

i-am-sijia commented Dec 30, 2022

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Uh oh!

Conversation

ghost commented Oct 4, 2022• edited by ghost Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ghostDec 7, 2022• edited by ghost Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ghostDec 7, 2022• edited by ghost Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

i-am-sijia commented Dec 15, 2022

Uh oh!

i-am-sijia commented Dec 30, 2022

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Uh oh!

Conversation

ghost commented Oct 4, 2022•
edited by ghost
Loading

ghostDec 7, 2022•
edited by ghost
Loading

ghostDec 7, 2022•
edited by ghost
Loading

ghost commented Oct 4, 2022•
edited by ghost
Loading

ghostDec 7, 2022•
edited by ghost
Loading

ghostDec 7, 2022•
edited by ghost
Loading

ghost commented Oct 4, 2022•
edited by ghost
Loading

ghostDec 7, 2022•
edited by ghost
Loading

ghostDec 7, 2022•
edited by ghost
Loading