Analysis of
Location Choice for Opening New Restaurants in Chicago
Chicago is one of the most populous
cities in United States, with an 2018 estimated population of ~2.7 million.[1] It
is a central city for finance, commerce, industry, education, transportation
and attracts various people to work, and live there. Opening a small
business-like restaurant in Chicago will an attractive opportunity for a lot
people.
Here, I examine a hypothetical
situation to choose location for opening a new business in Chicago. To narrow
down the problem, I specifically focus on location choice for starting a new
Chinese restaurant. As a common sense, the location is an essential factor of
the success or failure of a business and need a comprehensive analysis. This study
will be potentially useful for small business owner who want to start a new
Chinese restaurant.
Data Description
The Foursquare API is used to
retrieve most common venues in each community area of Chicago.[2] The data will
be used to cluster different community areas to identify attractive locations.
The other data I used are mainly
from the data portal of city of Chicago (https://data.cityofchicago.org)
(1)
Boundaries - Community Areas (current)[3]
This data defines the boundaries of
77 community areas in Chicago. I will use this data to make visualization of
different data and clustering of different community areas
(2)
Census Data - Languages spoken in Chicago, 2008 –
2012 [4]
This census data focus of
population spoken language and will be used to identity populous areas where
Chinese are widely used.
(3)
Crimes - One year prior to present [5]
This data records crimes reported
are used to identify safer location to start new business.
Furthermore, I use Foursquare API
to retrieve most common venues in each community area of Chicago. Based on
this, a clustering is performed to shed light on potential good new Chinese
restaurant location.
Methodology
The whole analysis is done in Jupyter
Notebook and has been uploaded in the GitHub repository.
In the analysis, I use community
area as the basic unit to do the analysis. Chicago has 77 community areas. The community
area data is obtained from the census data mentioned above. The latitude and
Longitude data were obtained by using python Geocoder library. The combined
community area data frame is made as shown in Figure 1.
Figure 1
Community Area DataFrame
In this study, I use python folium
library to make the geographic visualization. In the choropleth map, the Chicago
community area geojson file is used to find the boundary of different community
areas.
When analyzing the Chinese-spoken
population, a choreopleth map is made from the Census Data - Languages spoken
in Chicago dataset shown in Figure 2. As we can see, most Chinese spoken
population are concentrated in certain community areas in central Chicago.
Figure 2
Chinese-spoken population distribution by community area
To study the crime incidents distribution, the crime dataset is first grouped by community area and then made into a choreopleth map. Using this data, we can avoid certain community area when choosing the new business location.
Figure 3 Crime incident distribution by community area
A further clustering was performed to
explore the location choice. I used the Foursquare API to explore the community
area and segment them. The venue limit is set to 100 and the radius is set to 500
m from the given latitude and longitude of each community area. Then one-hot
key encoding is used to transform the data.
In the final clustering, the K-means
algorithm is used as it is the most common one and what I have learned from
this course. To simplify the analysis, I only use the Chinese Restaurant as the
feature for the clustering. The cluster number is set to 5 for simplicity.
Results
The main result of the clustering
is the map shown. Each dot represents a community area and the same color dot
are within the same cluster. The red color dot represents the cluster with no
Chinese restaurant presence according the Foursquare API. Similar to the Chinese-spoken
population, the Chinese restaurant are concentrated in some community areas.
Figure 4 Clustering of Chicago Community Area
Further, an intersection between cluster red community areas (with no Chinese restaurant presence) and the top 10 Chinese-spoken population community area was obtained: Mckinley Park, West Ridge, and Brighton Park.
A further check on the crime
incidents was performed on the three candidates are performed and all of them
are ranking in the middle the crime incidents lists. Therefore, all three make
the final recommendation list.
Discussion
Though Chicago is a diverse city,
the Chinese-spoken population are rather small and concentrated in community
areas like Bridgeport, Armour Square (where the Chicago Chinatown locates), McKinley
Park, and Brighton Park. In this analysis, I assumed that Chinese-spoken people
are more likely go to the Chinese restaurant. However, the leave out those who
love Chinese food but do not speak Chinese. It is to be decided if this will skew
the location analysis.
Crime issue is a concern for people
who live in Chicago. In the rough analysis using the total number of crime
incidents, I tried to avoid locations that rank high in the crime incident
number. In the future study, a crime incident in category (e.g. crime incidents
per restaurant by community area) are likely reveal more insights.
In this simply clustering of this
study, I only use the Chinese restaurant as the feature. Since competition
between different restaurant, a further study can be done with all the
restaurants or Asian restaurants.
Conclusion
Following the simple analysis and clustering,
I find Mckinley Park, West Ridge, and Brighton Park could be possible community
area to open a new Chinese restaurant in Chicago based on Chinese-spoken population,
Chinese restaurant clustering, and crime incidents. In the real situation, more
factors should be considered, and a more complex analysis should be conducted
before the investor can decide.
References
