Joshua Cohen - Notes

[paradoxwalk]

Week 1

• Did not meet this week, began to research surrogate modeling

Week 2

• Attended update meeting and subteam discussion, gained overview of VIP

• Submitted preferences for team meeting

Week 3

• Began specialization into Mobility 1
• Researched Urbano, WalkScore, walkability metrics generally
• Observed research gap between observed pedestrian environment and pedestrian scoring

Week 4

• Began research on qualitative pedestrian performance metrics
• SPACES, PEDS, PEQI
• Spoke with Dr. Pérez Guzmán on viability of these models

[4:24 PM] Cohen, Joshua R Meeting Notes 2/6 3:30p

Feasibility issues with online survey for walkability priorities in communities QR codes on structures to get online results Make consideration for where the surveys are being placed to avoid sampling bias Questions must be generalizable to general population as opposed to individual neighborhoods Walkability metric that provides weights for different feature priorities Every metric has a given weight For each neighborhood, those weights will be different Surveys are conducted by neighborhood to update weights dynamically for CURRENT neighborhood Personalized definition of walkability Questions to answer: How to evaluate walkability of large geographic areas, entire cities? How to compare weights between regions, a proxy of needs of a community. Next steps:

Focus on factors we want to focus on within flowchart

Week 5

• Looked into dimensions of walkability with a focus on non-utilitarian (leisure) walking, found transversabiltiy, compactness, connectivity, and comfort. TBD: how to account for overlapping factors within PEI existing sub-indices, how to weigh additional indices
• Found multilevel regression model describing impact of certain pedestrian infrastructure on perceived walkability, noted elements to be included in enhanced PEI. May inform weighing of certain pedestrian features

Week 6 Progress meeting notes:

• Began working on PEI incorporation into Rhino
• Starting with population density index (PDI)
• Working within global human settlement layer (GHSL) for population data within 100m cells
• Integrating this data with osmnx regional info for population per area

[paradoxwalk]

Weekly Notebook Entry — Week 7

Overview

• Week Span: 2/19 to 2/25

Tasks for This Week

List your specific objectives for this week, including learning targets and project milestones.

• [ ] Task 1: Determine how to use osm-population-predictor training model for feature-population relationship
• [ ] Task 2: Integrate osm-population-predictor with model to find population by arbitrary area

Tasks

Task: [Task Name] - (completed)

• Description:
• Link to Work: Link
• Outcome & Reflection:

Task: [Task Name] - (ongoing)

• Current Status:
• Expected Completion:

Challenges & Solutions

• Use PACE to train model on Atlanta OSM features with European Comission's GHSL (or Facebook's)
• How to migrate AWS model training to PACE?

General Takeaway, Skills & Knowledge Acquired

• How to use Anaconda (beginner)
• Methodology for producing PDI from arbitrary polygon on OSM (implementation pending)

Team Meetings

• Meeting Highlights: Change in methodology for Network Density Index, trying to use Google Places API. Open questions with calculation of Commercial Density Index, as should large commercial presences by floor area necessary correspond to greater walkability?
• Peer Feedback:

Personal Reflection

• Anaconda is a pain to set up.
• There's a tradeoff between population prediction accuracy and size of bounding region. Larger polygons produce more accurate estimations, vice versa for smaller. Whether using a geographical database like GHSL will prove more effective than a predictive model like osm population predictor is TBD by comparing samples of both to the zones investigated in Atlanta.

Visual Documentation (optional)

Additional Notes & Resources (optional)

• Resource Name - Brief description.

---

Plans for Next Week

• [ ] Next Week's Task 1: Description and strategy.

[paradoxwalk]

Weekly Notebook Entry — Week 8

Overview

• Week Span: 2/25 to 3/3

Tasks for This Week

• [x] Task 1: Write code to get census data from Census package
• [x] Task 2: Write code to get polygon outlines of census blocks/tracts using TigerWeb API
• [x] Task 3: Write code to get the building shapes within each tract, including area

Task 1 - (completed)

• Determine what data is available from the census api and what can be done
• Was able to successfully get required data for each census block group or tract.

Task 2 - (completed)

• Determine how to create polgyons of the census block group coordinates from TigerWeb API that match the coordinate description necessary for osmnx's features_from_polgon function
• Was able to successfully get the coordinates for each tract, tested and compared against satellite imagery to ensure accuracy.

Task 3 - (completed)

• Use the features_from_polygon function to get all of the features within a certain census group
• Was able to successfully get the data, although it's a bit slow when making the individual calls. Future effort will be for speeding up this process and extracting features more quickly. Perhaps by downloading the features of the county to begin as opposed to doing it on the fly?

Challenges & Solutions

• Finding out which API sources have which data; i.e., ACS from the Census has population data per tract but TigerWeb has the coordinates that define each tract

General Takeaway, Skills & Knowledge Acquired

• Interacting with TigerWeb API and US Census codes, using osmnx to extract features by polygon and converting between coordinate types (thank you for the help, Silvia!)

Team Meetings

• Meeting Highlights: Continue working on individual subindices, consider how some of the parameters of one indix will affect another (i.e. population density being a factor for the Commercial Density Index)

Personal Reflection

• I need to strengthen my python skills. Many of the roadblocks I'm facing are not due to not understanding the process, but rather the syntax of python. I feel like I've made great improvement since the beginning of the semester but I still need to look to the future to avoid 'rookie mistakes' that eat away at time.

Visual Documentation (optional)

Additional Notes & Resources (optional)

• Github code upload coming as soon as I clean it up a bit.

---

Plans for Next Week

• [ ] Next Week's Task 1: Make the PDI calculation a lot faster. Right now, it takes about 4 seconds to process each of the census blocks, which for Fulton County means 55 minutes of processing to get every census block.

[paradoxwalk]

Weekly Notebook Entry — Week 8

Overview

• Week Span: 3/4 to 3/10

Tasks for This Week

• [ ] Task 1: Write code for interpolation of population by census block groups within an arbitrary area
• [ ] Task 2: Determine how we want to divide area for evaluation with group
• [ ] Task 3 Learn how to use Jupyter notebook so my code is legible

Tasks

Task: [Task Name] - (completed)

• Description:
• Link to Work: Link
• Outcome & Reflection:

Task: 1 - (ongoing)

• Need to determine how arbitrary areas are defined; perhaps simply by block group? Would be most straightforward, 0 interpolation necessary

• Expected Completion: After team meeting, at latest during discussion with Patrick on Thursday

• Task: 2 - (ongoing)

• Component of Task 1, question of what is a reasonable expectation for a zone

• Expected Completion: At the same time as Task 1, they're directly related

Task: 3 - (ongoing)

• Asking the group to help teach me because I'm pretty clueless
• Expected Completion: By the end of the week, whenever one of my group members has the time to spare

Challenges & Solutions

• Come up with meaningful subdivision for walkability; what consitutes a reasonable area of investigation?
• How do I set up a Jupyter. What exactly is Jupyter? Why is Jupyter?

General Takeaway, Skills & Knowledge Acquired

Team Meetings

• Meeting Highlights:
• Peer Feedback:

Personal Reflection

Visual Documentation (optional)

Additional Notes & Resources (optional)

• Resource Name - Brief description.

---

Plans for Next Week

• [ ] Next Week's Task 1:

[paradoxwalk]

Weekly Notebook Entry — Week 10

Overview

• Week Span: 3/25 to 3/31

Tasks for This Week

List your specific objectives for this week, including learning targets and project milestones.

• [ ] Task 1: Finish shapefile generator by city, learn how to utilize US place codes, create integrated shape files for project

Tasks

Task: [Task Name] - (completed)

• Description:
• Link to Work: Link
• Outcome & Reflection:

Task: 1 - (ongoing)

• Current Status: Most of the code is written, but some place IDs don't work. Need to debug to determine where
• Expected Completion: End of week; if not able to complete, discuss manually generating shapefiles for city as opposed to automating

Challenges & Solutions

General Takeaway, Skills & Knowledge Acquired

Team Meetings

• Meeting Highlights:
• Peer Feedback:

Personal Reflection

Visual Documentation (optional)

Additional Notes & Resources (optional)

• Resource Name - Brief description.

---

Plans for Next Week

• [ ] Next Week's Task 1: Description and strategy.

[paradoxwalk]

Weekly Notebook Entry — Week 11

Overview

• Week Span: 4/1 to 4/7

Tasks for This Week

• [ ] Task 1: TBA after Monday meeting

Tasks

Task: Generate Place Shapefiles - (completed)

• Description: Generate shapefiles for region based on Place GEOID
• Outcome & Reflection: Temporarily haulted. The GEOID retriever is challenging to use as the Place IDs are often different from accepted city names. A shared shapefile library will instead be used to coordinate.

Task: [Task Name] - (ongoing)

• Current Status:
• Expected Completion:

Challenges & Solutions

General Takeaway, Skills & Knowledge Acquired

Team Meetings

• Meeting Highlights:
• Peer Feedback:

Personal Reflection

Visual Documentation (optional)

Additional Notes & Resources (optional)

• Resource Name - Brief description.

---

Plans for Next Week

• [ ] Next Week's Task 1: Description and strategy.

[paradoxwalk]

Weekly Notebook Entry — Week 13

Overview

• Week Span: 4/15 to 4/21

Tasks for This Week

• [ ] Task 1: Work with Neha on finalizing CDI
• [ ] Task 2: Finish sections of presentation on PEI
• [ ] Task 3: Look into tiling for PEI calculation

Tasks

Task: [Task Name] - (completed)

• Description:
• Link to Work: Link
• Outcome & Reflection:

Task: 1 - (ongoing)

• Current Status: Meeting on Friday to finalize the subindex
• Expected Completion: Friday evening

Task: 2- (ongoing)

• Current Status: Set up slides, created graphs used for demonstration
• Expected Completion: Draft completed by Friday, work over weekend

Task: 3 - (ongoing)

• Current Status: Wednesday review API and potential implementation with PEI scores
• Expected Completion: Friday meeting with Silvia

Challenges & Solutions

General Takeaway, Skills & Knowledge Acquired

• Tiling API

Team Meetings

• Meeting Highlights: Discussed outline for presentation
• Peer Feedback:

Personal Reflection

• Determining the appropriate boundary for each PEI calculation is a crucial step in the process

Visual Documentation (optional)

Additional Notes & Resources (optional)

• Resource Name - Brief description.

---

Plans for Next Week

• [ ] Next Week's Task 1: Description and strategy.