Group member: Huanghe YaoJing, Minxue Gu, Jinpu Cao
My task mainly focused on the analysis of " Negative utility" and the comparisons across time

The “complete communities” methodology was applied into a sub-geography: Menlo Park City in the Bay Area. We use blocks as our “origins.” The following mapping shows the 417 blocks in the city in 2020.

Isochrones

We analyze three modes of travel: walking, driving and public transportation.

Walking and Driving Accessibility

The following mapping shows the isochrones for walking and driving for 5 minutes, 10 minutes, and 15 minutes, from the centroid of each block in the city.

Public Transit Accessibility

Here, the public transit accessibility means walking and public transit accessibility. We use the algorithm mentioned in the class notes to implement. Specifically, filter out the stations outside the 15-minute walking isochrones, which are starting from centroids of each block of the city. Then find all stops the transit can arrive within 15 minutes (also including the walking time). Finally, draw the walking isochrones withing the left time starting from the final stops. The following mapping shows all the transit stops which are used to draw the transit isochrones.

The following mapping shows the transit isochrones for 5 minutes, 10 minutes, and 15 minutes, from the centroid of each block in the city.

Merge all the isochrones together. We get the following mapping.

POI - Complete Scores

We choose park, doctors, restaurant, school and supermarket as our critical amenities.

Amenity Preference Decay Model

We use exponential functions to model the “decay” of amenity value: the reduction in value for each additional amenity of a POI type, and the reduction in value for a trip the longer it takes. The following table shows the detailed parameters in our “amenity preference decay model.”

Amenity Preference
amenity amenity_value amenity_quantity amenity_decay
park 0.60 2 0.3465736
doctors 0.50 3 0.2310491
restaurant 0.25 30 0.0231049
school 1.00 1 0.6931472
supermarket 0.70 2 0.3465736

Travel Mode Preference Decay Model

We also use exponential functions to model the “decay” of travel mode preference. Besides, NHTS data is used to better calibrate the dacay index of the function. Our calibration is based on the following assumption: because half of trips were over this length, therefore only half of the population was willing to make trips over this length, therefore trips over this length are only half as valuable. In addition, there is no information related to transit time. We decided to assign 25 minutes as a reasonable time, which means the mode preference will decay to 0.5 when it will take 25 minutes to somewhere. The following table shows the detailed parameters in our “travel mode preference decay model.”

Mode Preference
mode mode_value mode_reasonable amenity purpose_label mode_decay
transit 0.5 25.00 school school 0.0277259
transit 0.5 25.00 restaurant restaurant 0.0277259
transit 0.5 25.00 doctors doctors 0.0277259
transit 0.5 25.00 park park 0.0277259
transit 0.5 25.00 supermarket supermarket 0.0277259
walking 1.0 10.00 school 8. Attend school as a student 0.0693147
driving 0.6 15.00 school 8. Attend school as a student 0.0462098
walking 1.0 10.00 supermarket 11. Buy goods (groceries, clothes, appliances, gas) 0.0693147
driving 0.6 10.00 supermarket 11. Buy goods (groceries, clothes, appliances, gas) 0.0693147
walking 1.0 9.50 restaurant 13. Buy meals (go out for a meal, snack, carry-out) 0.0729629
driving 0.6 15.00 restaurant 13. Buy meals (go out for a meal, snack, carry-out) 0.0462098
walking 1.0 11.00 park 15. Recreational activities (visit parks, movies, bars, museums) 0.0630134
driving 0.6 21.00 park 15. Recreational activities (visit parks, movies, bars, museums) 0.0330070
walking 1.0 13.75 doctors 18. Health care visit (medical, dental, therapy) 0.0504107
driving 0.6 15.00 doctors 18. Health care visit (medical, dental, therapy) 0.0462098

Complete Scores (2020)

Baseline complete score is defined by the sum of each decayed travel-amenity value when the quantity of each amenity is just equal to the half-value quantity (value decay to 0.5). Divide each block’s complete score by the baseline score and we can get the following mapping:

## Spherical geometry (s2) switched off

Critical Amenity Analysis

Choose the supermarket as the critical amenity. If there is a supermarket within the 15-walking isochrones, we define the block have the minimum access. The following mapping shows the distribution of “critical amenity accessibility” in the Menlo Park City.

Negative Utility POIs

After checking 129 types of amenities, we decided to choose fast_food and shelter as negative utilities. As far as fast food restaurants are concerned, we must all have been too busy to have time to eat. Therefore, there are a certain number of fast food restaurants in the community that can provide convenience to residents. However, too many fast food restaurants in a community mean that people here regard fast food as a common meal rather than a standby. Many studies have shown that fast food have a negative impact on human health, such as obesity (Pereira et al. 2005), diabetes (Pan, Malik, and Hu 2012), cardiometabolic disorders (Bahadoran, Mirmiran, and Azizi 2015), etc. Based on the facts above, the report assumes that the negative impact of fast food restaurants will increase quadratically with the increase of the number of fast food restaurants. Suppose the negative impact of the 1st fast food restaurant is only 1 and at this time, the growth rate is the smallest (0). Besides, assume the negative impact of the 10th fast food restaurant reaches to 2. That is to say: \[ Negative\ Impact\ of\ Fast\ Food = 0.01 \times (quantity\ of\ fast\ food) ^2 + 1 \] Similarly, in terms of shelter, it is hard to make a community without any shelters. But the quantity of shelters should be controlled in some range in order to provide residents a good public security environment. The report uses the similar quadratic curve to depict the negative impact of shelters. Here, suppose the negative impact of the 1st shelter is only 1 and the negative impact of the 20th shelter reaches to 2. That is to say, \[ Negative\ Impact\ of\ Shelter = 0.0025 \times (quantity\ of\ shelter) ^2 + 1 \] The following chart shows the negative impact of both fast food restaurant and shelter.

The following table shows the details of the report’s assumption about the amenity negative impact.

Amenity Negative Impact
amenity amenity_value amenity_quantity amenity_increase
fast_food -0.10 10 0.0100
shelter -0.05 20 0.0025

For travel mode, same as the demo in class, the report focuses on the travel mode: walking and driving. The decay mode is also exponential. Here, decay mode is called other than increase mode like the above since for negative utilities, the negative impact will decay with travel distance/time increasing. Similarly, the report uses the following mode_reasonable to calibrate the decay index.

Mode Preference
mode mode_value mode_reasonable mode_decay
walking 1.0 20 0.0346574
driving 0.6 30 0.0231049

Finally, we can get the negative utility score because of fast food and shelter in each blocks of Menlo Park. Since almost all the score are higher than baseline, relative score is used for mapping. From the mapping we can see that in the middle area of Menlo Park City, the negative utility caused by fast food and shelter is serious. Especially for the following 5 blocks, their negative utility scores are too high, which should be paied attentioned to.
\[Relative\ Score = 1 - \frac{Score\ - min(Score)}{max(Score)\ - min(Score)}\]

Five Blocks with Highest Negative Utility Score
baseline score = -1.29
GEOID20 total relative_score
060816139001005 -103.91049 1.0000000
060816139001008 -103.90533 0.9999498
060816117002008 -103.01119 0.9912574
060816117001000 -98.90170 0.9513071
060816139001004 -96.73408 0.9302346

Equity Analysis

In order to get a general picture of level of amenities of each block among different blocks in the city, We divide the complete scores into three types as the following table shows.

Complete.Score Level
< 1.5 Poor
1.5 ~ 2 Acceptable
> 2 Good

From the following mapping, we can see that most blocks in the city, the number and the distribution of the five amenities are acceptable. However, in northeast and southwest of the city, some blocks does not have enough amenities (or not close enough), which should be paid attention to.

The following chart shows the equality analysis of the complete score among several races. From the chart we can see White people tend to live in the area with more ‘Good’ amenities (more than their population ratio).

Hypothetical Demonstration

We find that the score of area near the East Palo Alto City is relatively low. Therefore, three new restaurants and a new supermarket are added. Then we can get the new complete score in the city.

From the mapping we can see that the complete score in that small area increases compared to the previous scores.

Cross Time Analysis

The following mapping shows the complete scores of each blocks in Menlo Park City from 2019 to 2022. From the mapping, we can see that the blocks in the center area of the city have the greatest complete scores and during the four years, their scores increased more quickly than other blocks’. From the view of urban development planner, the blocks of other area should also be considered when some new amenities are built.

The following chart shows the summary of complete scores of each blocks in the city decreased year by year. From 2019 to 2020, the decline is very obvious since the number of blocks increased from 2019 to 2020. In generally, new amenities are becoming more and more concentrated in some areas. In this case, the complete score will not increase obviously even decrease since the exponential decay will decay slowly with the increase of amenities.

From our analysis we think that the methodology of “complete score” is very useful when analyzing whether an area has convenient enough amenities. For urban planners, it can also be used to decide where the next amenity should be built. I think the “decay” part of this methodology is very interesting. This section might be subjective (one of the disadvantages of the method). But we can do many different assumptions in this section too. That is the beauty of the analysis.

Bahadoran, Zahra, Parvin Mirmiran, and Fereidoun Azizi. 2015. “Fast Food Pattern and Cardiometabolic Disorders: A Review of Current Studies.” Health Promotion Perspectives 5 (4): 231.
Pan, An, Vasanti S Malik, and Frank B Hu. 2012. “Exporting Diabetes Mellitus to Asia: The Impact of Western-Style Fast Food.” Circulation. Am Heart Assoc.
Pereira, Mark A, Alex I Kartashov, Cara B Ebbeling, Linda Van Horn, Martha L Slattery, David R Jacobs Jr, and David S Ludwig. 2005. “Fast-Food Habits, Weight Gain, and Insulin Resistance (the CARDIA Study): 15-Year Prospective Analysis.” The Lancet 365 (9453): 36–42.