bellevuewa / bkrcast Goto Github PK

Presently BKRCast (same as SoundCast) is building shadow prices at the end after feedback loops are done. Also, update DaySim sample rates to use 100% sample for all shadow price runs and feedback loops.

update setup after a calibration round (as of 06/23/2017)

The latest DaySim, checked out on 10/28/2016, gets stuck in running choice models. Apparently it is running into an infinite loop in destination choice. The older DaySim, checked out on 08/12/2016, runs fine with the same inputs.

Organize input configuration settings by putting all sub-modal settings into one file: input_configuration.py. Also, try to minimize hard coded input settings in scripts. This will help identify duplicate input file paths. Having one configuration file will be useful when an input setting needs to be updated, user would need to go to only one configuration file.

In the end, there will be three configuration settings:

1. input_configuration.py
2. emme_configuration.py
3. daysim_configuration_template.properties

We found our model did not toll SR520 and further investigation showed all model extra attributes were not imported. I think this piece could be missing in the network import. Can you double check and confirm? If it is missing, we need to put it in and figure out a way that allows us to import a number of extra attributes (node, link, turn, and other types) and makes it flexible (we can add extra attributes down the road without change code. We are ok to change configuration file).

@dong-bellevue

add Simpleinjector.dll and simpleinjector.xml to the daysim folder.

On @hu's machine, the Python scripts can't find other scripts in the current working directory. This use to work for him and still works for Nagendra and PSRC. Update the scripts by appending the current working directory to all scripts so it works.

"@stoll, @htoll, @Ltoll"

@s-yan

Create a Scenarios: a more thorough review of the Model Setting and Script files will help identify additional scenario types to make this section more complete

The special generator hard coded in emme_configuration.py is not a good implementation. It is ideal to save the special generators in an external file.

update setup to four time periods

Need to design a better way to update this emme_attr.in. So far this input file is not fully compatible with network and it needs manual update.

@dong-bellevue: We need to add some code to import zone partitions. Attached is the partition file. Now it has only two partitions but will have more later. It is not urgent but we do not want to miss this piece.

bike vols are assigned to all four time periods. But the bike weight (bkwt,in) calculated and exported in calc_bike_weight() only generates one set weight. Need to confirm on which time period @bkwt is calculated. Or make sure it is generated in AM time period.

Bellevue (Shuming and Hu) wants to assign a negative value to the coefficient 6. Presently the value is 1.10241152106. Update it to -1.10241152106.

@s-yan

Model Run Settings, provide more detail documentation of the three configuration files:
input_configuration.py,
emme_configuration.py, and
daysim_configuration_template.properties

• For each of the files, add an overview section to describe all the major functions the file accomplishes
• Perhaps combine this section with the sub-section called “Scripts”?
• Each of the Python file includes some documentation. Somewhere in the model documentation, please more fully explain the available option choices for each line in the sequence as they appear in the corresponding Python file, what each choice does, and when to select a certain option.

So far the power point file is for V1.2 (RSG's version).

It tries to reference out_lu_summary but it is not defined. Need to review and revise summarize_land_use_inputs.py.

@s-yan - It will be helpful to include a master directory diagram/tree to show how it looks like after set up, perhaps combine this section with the sub-section called “Master Directory”?

@s-yan

Utilities: The folder “psrc_to_bkrcast_scripts” includes several python scripts that are used to convert SoundCast data to BKRCast. How are these scripts implemented in the model run stream or where are the command/s that execute these scripts embedded? There are instances where some parameters in these files need to be updated/changed, describe when and how. (This again is tied to the Creating Scenario section).

Latest DaySim executable fixes population expansion factor bug

update the configuration file template to reflect updated coefficient file names. the latest calibrated PSRC coefficient files are being used now.

So far the tour mode share calculator only calculator all purposes and work. Need to separate other purposes as well. Put notes in between.

BKR is getting error while running popsampler.

update setup at the end of the project.

The PM peak hour model volume is 2000 vph (bidirectional), while the ADT is 30K. When the production/attraction at this station is tripled, the pm peak volume is increased to 5000 vph.

Possible solution is to increase external to external trip table.

update scripts for running accessibility calculations (parcel buffering) and daysim vs survey summaries

@s-yan

• Land use files: provide updated field definitions for the “parcel file”.
• Supplemental trip files: explain how and when they are converted from the PSRC’s 4K zone system to the BKR zone system

add popsampler.py to the model setup

New time period definitions:
am: 6am to9am
md: 9am to 3:30pm
pm: 3:30pm to 6:30pm
ni: 6:30pm to 6am

add the preferred sampling plan (BKR 400% and outside 21%) based on the first round of stochastic runs.

In the survey file, the transit pass field (“ptpass”) in the person data ranges from -1 to 6. Where,

1. -1 is “missing” (they didn’t ask that question on the university survey, but a lot of students have to buy a transit pass as part of their fees, so most of them probably have a pass)
2. 0 is no pass
3. 1-6 are various types of passes

The third category should be treated as 1 (yes) during calibration summaries.

update the setup to take extra attributes inputs from the user and add them during network import.

Specify an external location of the parcel file in the configuration settings and delete all copies of parcel files from the working directory after the model run is finished.

in accessibility.py, Line 172 throws an error:
transit_df = pd.DataFrame.from_csv(transit_stops_name, index_col = False)

Update "index_col" to "None". So, the new code:
transit_df = pd.DataFrame.from_csv(transit_stops_name, index_col = None)

So far we need to manually open each databank to copy the link and transit table to the validation spreadsheet. It will be nice to export the tables (AM, MD, PM, NI) to csv file before the BKRCast run is done.

The ESD parcel file being used by Hu has some field names capitalized but they are smaller case in LOADS parcel file.

Add code to capitalize field names to avoid this error:
parcels.columns = [i.upper() for i in parcels.columns]

Need to review and revise daily_bank.py