Fast Food vs. Location
'Are fast food locations and poverty correlated in the Los Angeles area?'
Introduction Video:
Scope of Study:
- My project started with noticing the serious issue of obesity in the United States
- The study of analysis is the entire Los Angeles county -- also including Palos Verdes and Santa Monica
- Obesity rates for adults in the United States are exceeding 35% of the population
Body Mass Index Over 30+
Percentage of Obese Adults: 1990-2018
"Body Mass Index (BMI) is a person’s weight in kilograms divided by the square of height in meters. A high BMI can be an indicator of high body fatness." -- Centers for Disease Control and Prevention
New research challenges old assumptions about why poor Americans eat junk food
The Sources Used for Data:
Methodology:
First, I had to find the HUD Low-Income Housing Tracts and Layer the Fast Food locations on top of it
- I chose the top SIX fast food restaurants in America from the POI Factory
- McDonalds
- Taco Bell
- Dominos Pizza
- Burger King
- Subway
- Kentucky Fried Chicken
The First Maps I Created:
- I added in the HUD Low Income Housing Tracts with the Fast Food Locations
- The fast food restaurants came in an excel sheet, so I had to use the XY to Point tool on ArcGIS Pro to place the points onto the map (shown Below)
- The 'Standalone Tables' shown in the bottom left tab is what the data came in on (Excel Sheets)
HUD Low Income Housing Tracts with SIX Fast Food Locations
- Since it did not matter what the actual fast food restaurant the data point was, I combined all of the location layers to one and changed the symbology to make it more readable
All Fast Food Layers Linked Together
- I chose the 'Fork and Knife' for the symbology because this was the best symbol to show any food location
Case Study:
Global Moran's I Summary
- Positive Moran's Index
- Z-score was extremely high and positive
- Variance was very little
"Deviation of spatial neighbors from the mean with respect to deviation at a location (without neighbors)" --Orhun
Positive Z Value Meaning: "Clustered spatial pattern. Data does NOT exhibit random patterns. Patterns exist in data such as high values occurring near high values and vice versa." -Orhun
- This means that there is clustering for the fast food locations and low-income housing tracts. This shows a relationship between the locations of fast food restaurants and where the low-income neighborhoods are.
Global Moran's I for the data
Poisson GWR
***wifi wouldnt load map at my house***
I am using a Poisson (Count) GWR since the dependent variable is discrete and represents the number of occurrences.
"% deviance explained by the local model—This is a measure of goodness of fit and quantifies the performance of the local model (GWR)." -ArcGIS
- If we ran this analysis, we would want a closer value to 1.0. This is the "proportion of dependent variable variance accounted for by the regression model." - ArcGIS
Density-based Clustering:
This tool focuses on where spatial data is clustered (points) and where the data is sparse.
Density Based Clustering
The data shown above does not give strong enough evidence that the data is spatially clustered in some areas compated to others
Results:
(Maps all shown above)
Well... do fast food restaurants correlate to low-income areas?
- The total number of fast food restaurants within the Los Angeles County was 966
- I used the 'Select by Location' tool to see how many fast food restaurants were within each low-income polygon -- this came out to 571
- 571/966=.591 which is approximately 59%.
- This means that 59% of the fast food restaurants lie within the Low Income Housing Tract Polygons
- This is not sufficient enough to say that there is a strong correlation between fast food restaurants and low income housing areas
NO, THERE IS NOT A CORRELATION!
Conclusion:
(Maps all shown above)
59% of fast food locations lie within low income housing tracts
- 59% is more than half! This does mean that there is more than half of the fast food restaurants lie within the low-income housing tracts
- However, I was expecting even a higher number such as 70% or even 80%!
- While it is true that the majority of fast food locations are in the low-income housing tracts, I will have to reject the null hypothesis because this number is not significant enough.
Future Findings:
If I had more time, this is what I would do...
- Find more fast food locations. Possibly go for a drive through Los Angeles. Make a list of all of the fast food restaurants I am seeing and write them all down on a list. Six fast food places is a good start, however, if I want to get more accurate, I aim to use upwards of twenty fast food restaurants that are in the Los Angeles Area (In-N-Out?!).
- I would research other possibilities of why fast food restaurants are placed where they are located. For instance, maybe I need to find zoning maps because perhaps the fast food restaurants cannot be built next to housing developments. Maybe I need to consider the cultural area that the community lives in. If it is China Town LA, are there going to be many more China-Wok fast food locations rather than Manhattan Beach? Probably.
- I think it would have been interesting to compare these results to a more rural area in California. I would assume that there are more fast food locations in large metropolitan areas such as Los Angeles compared to the highway through the desert going to Las Vegas. It would be interesting to see the correlation between fast food locations and income levels within rural areas.
- Lastly, instead of only using Low Income Housing tracts, I would like to see all levels of income. Possibly there are just as many fast food restaurants in low-income areas as there are in high and well-off areas in Los Angeles.
Acknowledgment:
This Story Map is created to satisfy final research project requirement for SSCI 381: Statistics for Spatial Sciences taught by Prof. Orhun Aydin.
