We should print just one message about file removals.

Emme writes directory names with an uppercase first letter, and case-sensitive operating systems (Debian) do not understand that database folder is the same as Database. The solution is to rename database to Database.

We need to add distance cost to car cost matrix in MockAssignment.

Transit vehicle kms and times are different in Helmet 3.1 and 4.0 CBA calculations even if using exact same network files. Difference in times seems natural, but transit kms should be same.

I'm not sure which method is correct. Test cases show that Helmet 3.1 prints more transit vehicle kilometres and therefore greater impacts than Helmet 4.0.

This is minor issue, but good to check at some point.

When running 10 iterations, the model seems to be running ok, but I get the following error between assignments, starting in later iterations of the model (in this example starting in iteration 9):

Traceback (most recent call last):
File "C:\Program Files\INRO\Emme\Emme 4\Emme-4.3.6\python27\lib\logging\handlers.py", line 77, in emit
self.doRollover()
File "C:\Program Files\INRO\Emme\Emme 4\Emme-4.3.6\python27\lib\logging\handlers.py", line 350, in doRollover
os.rename(self.baseFilename, dfn)
WindowsError: [Error 32] The process cannot access the file because it is being used by another process
Logged from file log.py, line 61

See file for an example of model output to command prompt:
Virheilmot_10_kierrosta.txt

Here is the console log. Sorry for the screenshot, copying text from XTerm proved to be too difficult... 😄 My Emme version is 4.3.5 and its Python version is 2.7.11.

In some cases there is a possibility of dividing integers, which may cause problems e.g. on following lines

The assignment crashes in the first iteration (Helmet 4.0.2) because the shape of cost_peripheral matrix is wrong. Matrix check is passed, and warning of additional zones is given. I wonder if this happened just for me?

Some python scripts handle the zonedata (.pop, wrk., etc) based on the order of vectors instead of reading the zone number from first column. I think the issue occurs in freight.py and in some trip generation scripts.

Usually pop and wrk files are modified in projects and wrong order may give unexpected results without warning.

There are two classes called GenerationModel:

Super confusing!

This might cause confusion if for some reason congested assignment is not used in end assignment

One more transit segment extra attribute has been added in Emme 4.4 for congested transit assignment.

Should remove the shape parameter and update return description accordingly.

Need to divide script into several

Not sure if this is an issue, but at least worth noticing in model development and when using MockAssignment.

When using MockAssignment model we cannot directly utilize results from previous model runs,. The transit impedance matrices saved in result_path/matrices folder are saved after congested assignment and therefore different from impedance matrices used in destination and mode share choice.
=> Model goes trough iteration, but mode shares and demand are considerably different from previous due to the transit impedance changes.

It would be useful to have a way to configure logging differently when running scripts via EMME (emme logging) and externally (F.ex during development, tests, etc).

We could use Python Abstract base classes for this purpose and have separate implementations for basic system.out-logging (perhaps to external file) and via EMME.

I was wondering if base demand car assignment should be done before transit cost calculation? Currently the transit assignment of cost calculations are done without any previous car assignments and speed of buses is dependent of link speeds produced by car assignment. I think this sets bus travel times infinite or zero, which could affect route choice.

However, I'm not sure how transit cost assignments is working and if this is a bug, so I won't create PR before comments.

OMX library is missing so cannot run integration-tests

Perhaps this file could also be in repository UI.

Atte Supponen suggested in his e-mail in Wed 19.6.2019 9:52 that summary results e.g. comparison of modal split in some areas could be useful in checking the validity of model run. These results could be presented as business graphic in addition to number tables.

At the moment, only generalized (weighted) travel time matrices are saved from transit assignment. For some analyses, trip parts and unweighted travel time is useful as well.

Base year should be changed to 2018 here:

Not an issue, but a wish, that total travel time (actual time) matrix is saved while trip_part matrices are saved. Would be nice if saved in omx-matrices since the matrice often used in evaluations.

Could it be wise to reserve the file workplaces.txt for the total number of workplaces and shares of total number of workplaces only and store parking costs in a separate new file? Maybe in the future there will be a need for new shares i.e. new columns in the former file.

Could it be wise to add a new column for parking costs in residential (i.e. generation) areas in the latter file?

This is probably just me but I am writing this here as a reminder!

Helmet run crashes when using the small test network. It is not compatible with congested transit assignment. Here is the error log:

2020-01-09 12:15:02,338 [INFO] Starting iteration 5
2020-01-09 12:15:02,351 [DEBUG] Added demand for freight, truck
2020-01-09 12:15:02,352 [DEBUG] Added demand for freight, trailer_truck
2020-01-09 12:15:04,053 [DEBUG] Added demand for hw, car
2020-01-09 12:15:04,055 [DEBUG] Added demand for hw, transit
2020-01-09 12:15:04,055 [DEBUG] Added demand for hw, bike
2020-01-09 12:15:04,391 [DEBUG] Added demand for hc, car
...
2020-01-09 12:15:07,664 [DEBUG] Added demand for oop, transit
2020-01-09 12:15:07,678 [WARNING] 1x1 matrix not matching 28 demand shares.Resorted to backup demand shares.
2020-01-09 12:15:07,678 [WARNING] 1x1 matrix not matching 28 demand shares.Resorted to backup demand shares.
2020-01-09 12:15:07,678 [DEBUG] Added demand for external, car
2020-01-09 12:15:07,710 [DEBUG] Added demand for external, transit
2020-01-09 12:15:07,717 [DEBUG] Added demand for external, truck
2020-01-09 12:15:07,733 [DEBUG] Added demand for external, trailer_truck
2020-01-09 12:15:07,733 [INFO] Assignment starts...
2020-01-09 12:15:07,963 [INFO] Car assignment started...
2020-01-09 12:15:08,394 [INFO] Car assignment performed for scenario 23
2020-01-09 12:15:08,723 [INFO] Calculates cumulative travel times for scenario 23
2020-01-09 12:15:09,549 [INFO] Congested transit assignment started
2020-01-09 12:15:10,963 [ERROR] Exception at iteration 5
Traceback (most recent call last):
  File "C:\Users\xxx\Home\helmet\helmet-model-system\Scripts\helmet_app.py", line 60, in run
    impedance = self.model.run(impedance, is_last_iteration=True)
  File "C:\Users\xxx\Home\helmet\helmet-model-system\Scripts\modelsystem.py", line 150, in run
    self.ass_model.assign(tp, self.dtm.demand[tp], is_last_iteration)
  File "C:\Users\xxx\Home\helmet\helmet-model-system\Scripts\assignment\emme_assignment.py", line 56, in assign
    self._assign_congested_transit(scen_id)
  File "C:\Users\xxx\Home\helmet\helmet-model-system\Scripts\assignment\emme_assignment.py", line 772, in _assign_congested_transit
    log_worksheets=False, scenario=scenario)
  File "standard\inro\emme\transit_assignment\congested_transit_assignment.py", line 583, in __call__
  File "standard\inro\emme\transit_assignment\congested_transit_assignment.py", line 636, in run_congested_transit_assignment
  File "inro\director\logging\trail.py", line 271, in __trace
  File "standard\inro\emme\transit_assignment\congested_transit_assignment.py", line 1337, in line_search
ArgumentError: Congested transit assignment cannot be applied to this transit network, please use Capacitated transit assignment instead.
2020-01-09 12:15:10,973 [ERROR] Fatal error occured, simulation aborted.
2020-01-09 12:15:10,973 [INFO] Simulation ended.

There is also something saying [WARNING] 1x1 matrix not matching 28 demand shares.Resorted to backup demand shares there.

Indata validation should included matrices.

Should maybe be done only in last iteration.

If dev-config has USE_FIXED_TRANSIT_COST set to true and UI false the model run is performed with fixed cost. I think this is unwanted behavior and selections of UI should override the dev-config file defaults.

The model system never uses this scenario, so eventually it crashes when trying to access matrices in Emme 4.4

This macro uses extra attribute @lvari:

But extra attributes are not initialized until here:

Are peak hour (aht, pt, iht) results needed in Excel Workbook, as a background information or other reasons, or could we expand them to 24 hours in cba.py with param.volume_factors. Writing to file and printing would be a bit easier when aggregated to single time period.

Previously, batin_4verkkoa_M2019.mac had some options to not run f-macros etc, which are missing from new 4verkkoa_HM40.mac. The old macro also did some basic validation of networks. Should these be re-added?

Fix needs to take into account that in-vehicle distance is used in calc_transit_cost, so this should not change into total distance.

Would need information from @zptro to figure how it is intended to be.

Part what fails every few runs:

else succeeding (with no changes, just a re-run):

Specific part what breaks:

For the mock assignment to work without emme installation, in ´helmet.py´, we can not import EmmeProject straight but we have to something like:

try:
from emme_bindings.emme_project import EmmeProject
except:
pass

Somehow this did not come up earlier when testing #244. Last iteration of EmmeAssignment returns error NameError: global name 'pandas' is not defined.

In addition to total number of residents in zones and shares of population of different ages there are also following columns in the file population.txt

• caruse: share of population that is car main user
• cardens: cars per inhabitants

When future years are concerned, are values in these columns supposed to be statistical input data or intermediate results of car use models?

Share of car users has an exception in two places which is not handled properly:

When population in the area is zero, tests will throw RuntimeWarning: invalid value encountered in double_scalars. It is a warning, not an error, so it will not affect the model run but it will show up when running integration tests. If it was not numpy matrices, it would probably throw a ZeroDivisionError.

Noticed when using a test network, and running test_models().

This stub was replaced with assignment.departure_time and should be removed.

Slots for car travel time are created in EmmeAssignment, but they are not used.

Base demand assignment (fixed cost = true) is crashing due to the matrix check is trying to find bike_work from cost OMX-file. Opening omx-files as MatrixData-object checks that all param.transport_classes are found and this crashes when base demand assignment is trying to read read fixed transit costs from file.

If I'm correct the cost_ttt .omx should not include bike_work?

IndexError: File C:\Users\YY\helmet\system-results\V2016\Matrices\cost_aht.omx does not contain bike_work matrix.

Python 2.7 is reaching EOL.

Helmet user interface crashes if previously defined "HELMET-skenaarioiden tallennuspolku" path doesn't exist anymore (e.g. folder deleted). As consequence, user interface shows empty screen without warning. Not sure if the problem occurs with other defined folders.

For calculations of zone specific mode choice, walk demand sum needs to be printed.

As discussed in #269, pylint is imported but not used. Here are steps to solve this:

• Look into tools to enforce good coding style
• Fix pylint messages
  • Documentation here: https://pylint.pycqa.org/en/1.9/technical_reference/features.html
  • Look into automated solutions for warnings regarding dull things like whitespaces
  • It may be beneficial to fix one message, report, category, or checker at a time. For example, the following throws only trailing-whitespace warnings: pipenv run python -m pylint --disable=all --enable=trailing-whitespace ..\Scripts\ This way the commits stay more informative.
• Could Travis throw a warning for unaccepted coding style?

Noticed that Emme-assignment methods auto_link_24h, transit_segment_24h and bike_link_24h are called separately for each time period as part of assignment (these methods also loop over time periods), which causes some extra calculations. This does not take much time, but is unnecessary anyways.

This can be easily fixed by calling these from modelsystem. I'll fix this after more urgent issues.