University of Southern California Dissertations and Theses

Operational optimization model for Hungry Marketplace using geographic information systems

(USC Thesis Other)

Operational optimization model for Hungry Marketplace using geographic information systems

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Copyright 2021 Mason Grant

Operational Optimization Model for Hungry Marketplace
Using Geographic Information Systems

By

Mason Grant

A Thesis Presented to the
FACULTY OF THE USC DORNSIFE COLLEGE OF LETTERS ARTS AND SCIENCES
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(GEOGRAPHIC INFORMATION SCIENCE AND TECHNOLOGY)

August 2021

ii

To Bill and Pam Grant, my Grandparents

iii
Table of Contents
Dedication …………………………………..………………………………………..….. ii
List of Tables …………………………………..……………………………………..….. v
List of Figures ……………………………………………………………………….…... vi
List of Abbreviations …..…………………………………………………………......… vii
Abstract ……………………..…………………………………………………………. viii
Chapter 1 Introduction …..…..…..……………………………………………………..… 1
1.1 Meal Delivery …..………………………………………………………….… 3
1.2 Study Area …..……………………………………………………………..… 4
1.3 Document Structure …..……………………………………………….…...… 6
Chapter 2 Background …..………………….…………………………………..……...… 8
2.1 Hungry’s Current Operations …..………………….……………..………...… 8
2.1.1 Operational Flow …………………………………………………... 8
2.1.2 Operating Locations …………………………………………….… 10
2.2 Literature Review …………………………………………………………… 11
2.2.1 Supply Chain ……………………………………………………… 12
2.2.2 Site Suitability …………………………………………………….. 13
2.2.3 Database Design …………………………………………………... 14
Chapter 3 Methodology ………………………………………………………………… 16
3.1 Data Collection and Processing ……………………………………………... 16
3.1.1 Offices ……………………………………………………………. 17
3.1.2 Chefs and Kitchens ……………………………………………….. 19
3.1.3 Clients …………………………………………………………….. 21

iv
3.1.4 Orders …………………………………………………………….. 22
3.2 Network Analysis …………………………………………………………… 24
3.2.1 Closest Facility ……………………………………………………. 25
3.2.2 Origin Destination Matrices ………………………………………. 27
3.3 Database Creation …………………………………………………………... 29
3.4 Operations Analysis ………………………………………………………… 30
3.4.1 Current Operations ………………………………………………... 30
3.4.2 Optimized Operations …………………………………………….. 31
Chapter 4 Results ……………………………………………………………………….. 34
4.1 Current Operations ………………………………………………………….. 35
4.2 Optimized Operations ………………………………………………………. 36
Chapter 5 Discussion and Conclusion …………………………………………………... 40
5.1 Assumptions and Limitations ……………………………………………….. 40
5.2 Conclusion ………………………………………………………………….. 43
References ……………………………………………………………………………… 45
Appendix A: Offices ……………………………………………………………………. 48
Appendix B: Kitchens …………………………………………………………………... 49
Appendix C: Clients …………………………………………………………………….. 50

v
List of Tables
Table 1 Offices Feature Class ......................................................................................................... 2
Table 2 Chefs Feature Class ........................................................................................................... 2
Table 3 Kitchens Feature Class ....................................................................................................... 2
Table 4 Clients Feature Class ......................................................................................................... 2
Table 5 Orders Feature Class .......................................................................................................... 2
Table 6 Clients ................................................................................................................................ 2
Table 7 Kitchens ............................................................................................................................. 2
Table 8 Office_Kitchens_OD ......................................................................................................... 2
Table 9 Kitchens_Clients_OD ........................................................................................................ 2
Table 10 Transportation Costs ........................................................................................................ 2
Table 11 Current Costs ................................................................................................................... 2
Table 12 Optimized Costs ............................................................................................................... 2
Table 13 Order Summary Statistics ................................................................................................ 2

vi
List of Figures
Figure 1 Hungry’s Operational Flow .............................................................................................. 2
Figure 2 Hungry's Current Operating Locations ............................................................................. 2
Figure 3 Methodology..................................................................................................................... 2
Figure 4 Office Locations ............................................................................................................... 2
Figure 5 Hungry's Kitchen to Client Orders ................................................................................... 2
Figure 6 Database ERD .................................................................................................................. 2
Figure 7 Current vs. Optimized Office Locations .......................................................................... 2
Figure 8 Current vs. Optimized Kitchen_Client_OD ..................................................................... 2

vii
List of Abbreviations
ERD Entity Relationship Diagram
GIS Geographic information system
SQL Structured Query Language
SSI Spatial Sciences Institute
USC University of Southern California
VRP Vehicle Routing Problem

viii
Abstract
The food industry has been completely disrupted over the past 5 years with the onset of
platforms like Doordash, Instacart, HelloFresh, and Hungry Marketplace. These platforms, and
others like it, offer customers timely, cost-saving, and convenient ways to prepare, consume,
and/or experience meals. The explosion in food-industry innovation has changed the dynamic of
food services altogether as status quo operations aren't meeting the needs of the innovative oper-
ations. More people are opting in to having their meals delivered to more comfortable, familiar,
locations as opposed to going to brick-and-mortar restaurants to consume their meals. As a re-
sult, the food industry’s changing dynamic has led many to reconsider owning expensive restau-
rant locations in consideration of alternative commercial locations that are much cheaper and al-
low them to deliver the same quality of service. Given the importance of kitchen and office loca-
tions and the critically changing nature of the industry, new methods are needed to determine the
optimal locations for companies that are delivering services in this new environment. Using a
GIS (Geographic Information System), this thesis uses spatial analysis including site-suitability
analysis and network analysis to build an optimization model for Hungry Marketplace, a food-
industry start-up. The model utilized the company’s current operations in Boston, Massachusetts
as a case study. The model identified optimal locations for kitchen and warehouse operations that
maximize the market opportunities while keeping the operational expenses low. This thesis pro-
vides recommendations to the company for a cost-effective operation going forward. Ultimately,
this data-driven and reproducible methodology can be applied by existing and potential compa-
nies in the food industry for optimizing their spatial decisions.

1
Chapter 1 Introduction
Food is not only necessary to sustain human life, but it is also the bedrock of civilization.
Throughout history, the act of eating food has extremely important social and cultural signifi-
cance. Humans have used the act of eating a meal for the most momentous occasions and some
of the greatest moments that have redefined history have been made during the act of having a
meal, including, the last supper, the first Thanksgiving dinner in America, Churchill and Stalin in
1942, and countless other undocumented dinners that led to revolutionary changes in society.
Although civilization has changed dramatically over these time periods, the cultural significance
has remained consistent. As the world continues to revolutionize the areas of science and tech-
nology, and more social dynamics become digital, it is more important now than ever that the
world takes the time to enjoy a fulfilling meal. Although technology may be pushing to remove
us from the present moment, it does not mean that we cannot also embrace technology to help
build a stronger social and cultural fabric that allows us to reinforce the human connection
through food.
Approximately 746 billion U.S. Dollars are spent each year at restaurants in the United
States alone (Lock 2018). Restaurant are the location of great meals, however, there is also
something to be said about the authenticity and comfort that is experienced when having a meal
in your own surroundings. These two situations are ideal, however, there are many factors like
time, cost, and convenience that deter people from these restaurant experiences and entice them
towards inferior alternatives. Using technology, many companies are looking to solve the prob-
lems of time, cost, and convenience to help more people enjoy a meal.

2
During the past several years, web and mobile technology has allowed companies like
Doordash, Hungry Marketplace, Instacart and others to redefine the food industry with innova-
tive approaches to how people consume meals (Oliver 2015). Doordash uses its platform and
network of drivers to pick up meals from restaurants around town and deliver them to individual
consumers wherever they are in under an hour. Similarly, Instacart uses its platform and network
of delivery vehicles, but its service is to pick up groceries from stores around town and deliver
them to individual consumers. Hungry Marketplace’s platform connects a large group of local
chefs that prepare meals and caters the prepared meals including delivery, setup, and teardown.
There are many other companies like the above three in the United States and throughout the
world that are focused on solving the problems of time, cost, and convenience that people in to-
day’s modern society struggle. All of these companies have the shared goal of allowing more
people to take the time to have a meal.
One commonality between all these food preparing/servicing/delivery companies is the
locational aspect that are necessary for successful operations. At a bare minimum, the companies
mentioned above, and other alike, must have a physical location to prepare and produce the food,
and another physical location to deliver the food to be consumed. There are additional locations
to take into consideration for these companies like office locations, and storage facilities for sup-
plies and vehicles. Generally, each of these companies encounters some transportation and logis-
tics costs given their locations, and if these locations are not optimized, it could destroy a com-
pany’s bottom line. The companies described above, and others like it, require some sort of spa-
tial decision making in order to begin operations. The quality of these spatial decisions could po-
tentially make or break the viability of their operations, which underscores the importance of a

3
thorough spatial analysis to help optimize location decisions and set the company up for success
from the beginning.
This thesis uses Hungry Marketplace, its data, and operational information as a case
study to build out the initial model that allows Hungry to make smarter decisions on their kitchen
and office locations. This model will allow Hungry Marketplace to identify the most optimal lo-
cations and create a framework for its operational guidelines that allows it to operate in an effi-
cient, cost-saving environment. Ultimately, the model uses a data-driven approach to provide a
replicable method for which stakeholders can confidently look to verify the decision information
and to make strategic business decisions moving forward.
Meal Delivery
The food-delivery industry is changing rapidly throughout the United States and the
world with no signs of stopping. In as little as five years, companies like Doordash, Postmates,
Blue Apron, HelloFresh, Instacart, Hungry Marketplace and many others have systematically
changed the way the food industry operate by delivering food-related services that save consum-
ers time and money. As these companies continue to grow and others continue to innovate, the
changing dynamic is introducing completely new ways of thinking into an industry that has been
around in some shape or form since the beginning of time. One of the cornerstones of the food
industry has always been and will continue to be a physical location to prepare the food. As these
new ways of thinking absorb into the minds of investors, new questions are being considered and
the status quo is being challenged. Seeing as location is one of the most important decisions a
company in the food industry can make, the location optimization that has been conducted in this
thesis will prove to be extremely valuable.

4
In 2016, a company called Hungry Marketplace launched in Arlington Virginia as an in-
novative meal-delivery company passionate about supporting local chefs and delivering meals
through its online platform. Today, Hungry has expanded its operations to seven different cities
throughout the United States and has partnered with hundreds of local chefs to prepare catered
food for customers within each market (Mahmoudpour 2020). Customers utilize Hungry’s online
platform and network of local chefs to order catered meals for events, team outings, or special
occasions. Once the order is placed, Hungry uses its team to deliver, set-up, and take-down the
meals at the customers desired location. There are at least three locations visited throughout
Hungry’s meal-delivery operation. First, the Hungry team has an office where their operations
are organized, and supplies are stored. Next, the Chef’s prepare meals out of commercial kitch-
ens that are located in various spots throughout a city. Finally, the customer choses their desired
location where they want the meals to be delivered.
Study Area
Hungry Marketplace began its operations in Arlington, Virginia in 2016 and, by 2019,
had expanded to six other cities in the United States with plans to open numerous other markets
throughout 2020 and beyond. Currently, Hungry operates in the D.C Metro Area; New York City
Metro Area; Atlanta, Georgia; Austin, Texas; Boston, Massachusetts; Dallas, Texas; and Phila-
delphia, Pennsylvania.
This study utilizes historical operating data collected from Hungry’s operations in Bos-
ton, Massachusetts. Hungry began serving the Boston market in June 2019 and continues to
serve it today with one office location, 20 full-time employees, and over 40 chefs. This study
uses Hungry’s order data in the Boston market during a nine-month period ranging from June 1,
2019 to March 31, 2020.

5
Boston, Massachusetts metro area is the tenth largest in the United States with a popula-
tion of 4.9 million people, spanning nearly 3,500 square miles, in 2018. The median age is 38
years old and per capita income is roughly $47,000 (Bureau 2018). Boston has played a signifi-
cant role in the United States and beyond with its powerful academic institutions including Har-
vard and MIT calling the historic town of Cambridge, home. Kendall Square, a neighborhood in
Cambridge, has been called “the most innovative square mile on the planet,” referencing the vast
amount of innovation in science and technology that has been born from public, private, and uni-
versity organizations in this area (WorldAtlas 2019). Given this innovation, Boston boasts a vast
network of highly educated business professionals working throughout the metro area. This de-
mographic is ideal for meal delivery as most people work demanding hours at innovative compa-
nies and enjoy the convenience offered by meal delivery.
As with most large metro areas in the United States, Boston suffers from transportation
congestion. In 2018, a study from INRIX found that Boston had the worst rush-hour traffic in the
United States with each driver, on average, wasting over 164 hours commuting each year
(INRIX 2020).Given the congestion in Boston, companies relying on transportation to offer their
service must take these concerns into account to find optimal locations that allow them to max-
imize their time and not waste precious dollars sitting in traffic.
Hungry Marketplace has capitalized on the enticing business innovation and dense popu-
lation of wealthy companies in Boston by offering quick, efficient, and tasty catered meals di-
rectly to company offices. This not only saves employees precious time, but it also allows com-
panies to offer benefits to their employees, enticing them to work harder and longer at the office.
Although Hungry has capitalized on the healthy market demand for corporate meal catering,
Hungry’s operations rely heavily on transportation throughout the city. Transportation costs have

6
not been thoroughly considered throughout Hungry’s current operations in Boston. Transporta-
tion costs could be severely impacting Hungry’s business, considerations that are analyzed fur-
ther in this study.
Document Structure
The remaining sections of this study include the background, methodology, results and
conclusion. Chapter 2 discusses Hungry Marketplace’s operational intricacies. Additionally, this
chapter discusses literature concerning site suitability and location optimization. Chapter 3 dives
deep into the methods used to conduct this study. Chapter 4 presents the results that were found
after conducting the methodology. Finally, Chapter 5 discusses the study’s limitations, future
work and ultimately concludes with recommendations for Hungry Marketplace’s operations in
Boston.
Chapter 2 first discusses how Hungry Marketplace operates. It details exactly what Hungry
Marketplace is and then presents Hungry’s operational flow which is the foundation for all meal
orders that are placed on its platform. Once Hungry’s operational groundwork is understood,
more detail is given into operations specific to the Boston market including current operating
locations and historical order data. The chapter then moves onto a literature review to provide
background and justification for the methodology proposed in the next chapter.
Chapter 3 presents the methodology used to guide the analysis in this study. The chapter
begins with a broad overview of the methodology, with subsequent sections diving much deeper
into the specific steps taken to achieve the desired analytical insights. Subsequent sections detail
considerations taken during data collection and processing, network analysis, database creation,

7
and operational analysis. The insights gathered from these sections fuel the results that are pre-
sented in the next chapter.
Chapter 4 discusses the results obtained from the methodology described in the previous
chapter. The results include detailed operational costs for Hungry’s current operations, which is
based on order data from June 2019 to March 2020. The results chapter also presents potential
cost savings if Hungry took steps to optimize its transportation costs through more efficient lo-
cation decisions.
Finally, Chapter 5 brings the study home by reflecting on its assumptions and limitations.
The chapter goes further into discussing how future work could build from this study to include
additional inputs that would make the insights gathered in Chapter 4 more robust. The chapter
then concludes the entire Thesis by summarizing its objectives, analysis, and results. This study’s
aim is to build a model that can definitively recommend operational improvements for Hungry
Marketplace’s delivery operations in Boston Massachusetts.

8
Chapter 2 Background
To provide context for this study, this section presents a detailed description of Hungry’s
current operations and a literature review to justify the study’s subsequent methodology. First,
Hungry’s operations are described by identifying its operational flow, operating locations, and
historical order data. Next, a literature review is conducted, outlining other site-suitability and
location optimization studies that have been performed.
Hung ry’s Current Operations
Hungry has created an online meal ordering platform for its clients. Hungry presents spe-
cific meal choices on its platform where clients can then order meals by specifying order details
including date and time, meal quantities, and more. Once a meal is ordered from the online plat-
form, Hungry uses its network of employees, vehicles, offices, and kitchens to prepare and de-
liver the meals to the client. Hungry has provided data on its operations from June 1, 2019 to
March 31, 2020. Using this data, the following sections present Hungry’s specific operational
flow, operating locations, and order history in the Boston market.
2.1.1 Operational Flow
Figure 1 shows Hungry’s basic transportation flow. A typical Hungry order begins at Hun-
gry’s office location where employees called “Captains” pick up the catering supplies needed to
set up the meals. Once the supplies have been obtained at Hungry’s office, Captains travel by car

9
from the office to one of the dozens of potential commercial kitchens where the meals could be
cooked from. Once the meals are picked up, Captains deliver the meals to the client’s location.
In Hungry’s current operating model, chefs are not employees but rather contractors that
are expected to maintain their own commercial kitchen. Most chef’s rent space from larger com-
mercial kitchens that offer monthly occupancy rates and are located throughout the metro area.
This results in each chef being locked into their kitchen location per their specified rental agree-
ment. In this typical situation, each chef is responsible for their own lease, kitchen equipment,
and supplies. Hungry is only responsible for paying the chef per order that is placed through the
online platform. Each chef has their own unique cuisine and style of meals that they cook from
their location. There is no standardization across the kitchen locations, therefore when a client
orders a meal, it will always be prepared by that particular chef in that particular location. Each
chef is unique to exactly one kitchen location, and Hungry simply presents their network of chefs
and meals to each client. The client is then able to choose whatever meal they would like to order
on Hungry’s online platform.
Once the client chooses a meal, the chef is notified with order details including date and
time the meals need to be ready, quantity of meals to be prepared, and any other requirements the
client has requested. Once the chef prepares the meals, Hungry’s Captain travels from the office
Figure 1 Hungry’s Operational Flow

10
to pick up the prepared meals. Once the meal has been picked up, the Captain transports the meal
to the Client’s specified location. Finally, Captains arrive at the client’s location where they setup
the meals, serve the meals, then tear down the meal set up and bring the supplies back to the
Hungry Office.
2.1.2 Operating Locations
Hungry currently operates out of one office in Downtown Boston, which can be seen in
Figure 2 below. The location was chosen in the early 2019 based on pre-launch partnerships, as
well as other considerations such as initial employee housing locations, and general business
suitability. Given the pre-launch status of the company, the decision to locate Hungry’s office in
Downtown Boston was not based on operational insights. Moving forward, the current office
location lease agreement expires in August 2020, and Hungry leadership is interested in relocat-
ing to another area that makes sense from an operational cost perspective.
Hungry has partnered with 37 chefs across the region. Some chefs operate out of the same
commercial kitchen location because the commercial kitchen companies offer multiple spaces
for individual chefs to lease. Given this, Hungry has delivered meals from 33 distinct kitchen
locations across the region, which can also be seen in Figure 2 below.

11
Since Hungry’s launch in Boston in 2019, the company has delivered 788 orders to over
100 clients throughout the region. Client locations can be seen in Figure 2 below.
Literature Review
On-demand meal delivery is a relatively new space in the food industry, therefor not much
literature has been published that specifically focuses on location and operational optimization
for on-demand meal delivery companies. However, many scholars have presented supply chain
models for the food industry (Bosona et al. 2013; Hekmatfar 2009). These models discuss loca-
tional considerations for the supply, demand, and fulfillment of food sources for traditional food
Figure 2 Hungry's Current Operating Locations

12
industry companies like restaurants and grocery stores. In addition to supply chain analysis, site
suitability analysis and database design are important considerations to build an operational op-
timization model. This literature review is divided into three sections that provide a greater un-
derstanding into related work that has been conducted for supply chain, site-suitability, and data-
base design. Supply chain literature is first discussed, which defines the considerations necessary
to find optimal locations for supply, demand, and fulfillment operations. Next, site-suitability
literature is reviewed, discussing the tools used in a GIS that will provide spatial insight into
various types of locations. Finally, database design literature is discussed which will inform the
best ways to create a scalable database that can be used for business purposes. This thesis aims
to recreate a similar methodology that takes these factors into account. This literature review
provides context for the methodology presented in the next chapter.
2.2.1. Supply Chain
The food industry relies on a vast network of local supply and fulfillment locations that
provide resources to companies providing a service to customer locations. On-demand meal de-
livery similarly relies on a supply chain network to deliver meals to customer locations. In the
study GIS-based Analysis of integrated Food Distribution Network in Local Food Supply Chain,
the authors discuss how local food producers need to integrate within networks of suppliers, dis-
tributors, and customers in order to increase their competitiveness (Bosona et al. 2013). The study
uses GIS to conduct location and route analyses to determine the most optimal supply chain con-
figuration to increase competitiveness for a company in Sweden. The study identifies that logis-
tics problems are the bottleneck in the local food systems and that location decisions are one of
the critical elements in strategic planning for different firms. The study considers supply and
fulfillment center locations and uses GIS to optimize these locations. The study also builds on

13
another study of location theory conducted by Farahani & Hekmatfar in 2009. Farahani & Hek-
matfar aimed to locate a single warehouse at an optimal location that minimizes the distance
between the warehouse and delivery points (customers) (Hekmatfar 2009). Both of these studies
determined that the main factors to be considered are the nature of the facility itself, environment
in which it is to be located, customers to be serviced by the facility, and a metric that indicated
distance and time between customers and facilities. The metric that indicates distance and time
between customers and facilities is presented in the next section discussing site suitability.
2.2.2. Site Suitability
In addition to the supply chain consideration required to find optimal operating locations,
site-suitability methods can also be used to calculate the suitability of these locations. The suita-
bility of a supply chain locations can be found by assigning each a pair of locations with a cost
associated with transporting good from one location to another. In Installing Public Electric Ve-
hicle Charging Stations: A Site Suitability Analysis in Los Angeles County, California, Jennifer
Jin uses site suitability analysis to determine suitable vehicle charging station location in Los
Angeles (Jin 2016). Although the study is not focused on the food industry, it uses GIS tools to
assign suitability costs to charging station locations. Similarly, this thesis uses similar tools to
assign costs to facility locations. In the book The Esri Guide to GIS Analysis Volume 3: Modeling
Suitability, Movement, and Interaction, Andy Mitchell presents many of the tools ArcGIS Pro
has to offer that were used to conduct this analysis (Mitchell 2012). The book gives context on
the understanding for site suitability, and for the Network Analysis section of this report. For the
network analysis section of this report, “finding suitable locations”, “rating suitable locations”,
and “modeling interaction” are sections of this book that will lay the foundation for how the
section was determined.

14
The tools used for this analysis include closest facility and origin destination matrices.
Both of these tools use the vehicle routing problem (VRP) method, which is one of the most
significant problems in the goods distribution management industry (Bosona et al. 2013). VRP
finds the most optimal routes in a distribution system where vehicles should be assigned to serve
a set of customers. When using VRP, important issues such as number of stages, fleet size, vehicle
capacity, delivery time window, and supply/demand nature need to be addressed (Giaglis 2004).
Although there are constraints such as location, vehicles, and paths between location, this thesis
relies on Hungry’s operational data, ArcGIS Pro and ESRIs road network to conduct VRP. The
VRP results show costs for a pair of locations in terms of time and distance. These costs will be
transformed into actual dollar amounts per minute and per mile using data from Hungry. Finally,
the costs associated between two locations are assigned, and locations are selected that minimizes
time and distance between a set of locations.
2.2.3. Database Design
The aim of this thesis is to discover insights from the data that was not only provided by
Hungry but also created throughout the methodology. The data gathered and created in this thesis
is not actionable by itself and needs to be collected and imported into a database not only to in-
form the results section of this thesis, but to also allow Hungry Marketplace to answer any re-
maining questions or to inform future work. In the USC Master’s Thesis titled Creating a Geo-
database and Web-GIS Map to Visualize Drone Legislation in the State of Maryland, Brendan
Blee shows that creating an effective database has several steps including determining what data
is needed, specifying the relationships and interactions in the data by creating an entity relation-
ship diagram (ERD), and arranging the data into a final database (Blee 2016). First, existing and
candidate supply, demand, office, and order data was collected and is presented in Section 3.1 of

15
this thesis. Additional data was created using network analysis and is presented in Section 3.2 of
this thesis. Next, the relationships and interactions were created in an ERD which is presented in
Section 3.3. Finally, the ERD was used to create the final database using Microsoft SQL Server.
The final database was created in a scalable way so that future operational insights could be
added if Hungry’s operations continue in the Boston market. The database allows users to query
information and gain insights into Hungry’s operational data which is ultimately stored in the Or-
ders table feature in section 3.1.4.

16
Chapter 3 Methodology
The goal of this study’s methodology is not only to analyze transportation costs of Hun-
gry’s current operations in Boston, but to also look at alternative operational locations that
could optimize its future operations. Hungry provided historical data to support current opera-
tional analysis which was presented in Chapter 2. Additionally, Hungry provided resources and
discussion for how they look at future developments, including candidate commercial kitchens
and office locations. Taking both existing and candidate location data into account, the method-
ology presented below outlines the workflow used to achieve a comprehensive review of Hun-
gry’s meal delivery operations in Boston, Massachusetts.
Figure 3 represents the general methodology and workflow that was followed. All analy-
sis was conducted using the WGS 1984 spatial reference system. There are four main compo-
nents to the study’s methodology which are expanded upon in much greater detail in subsequent
sections. First, existing and candidate location data was collected and processed to be used in
the network analysis section. The results from the network analysis are used to analyze trans-
portation costs for Hungry’s historical orders as well its recommended optimized operations.
Figure 3 Methodology

17
Data Collection and Processing
The first step of the methodology consists of collecting and processing the necessary data
to analyze Hungry’s meal delivery operations. As described previously in section 2.1.1 of this
study, a typical Hungry order begins at their office location where supplies are acquired by Hun-
gry Captains. Captains travel to each order’s specific kitchen to pick up the meals and then con-
tinue to the client’s location where they deliver their service. This operation consists of three
main locations: Offices, Kitchens, and Clients. Existing and candidate data for each of these lo-
cations was collected for the model. Additionally, Hungry provided a comprehensive dataset de-
tailing its orders from June 1, 2019 to March 31, 2020. Subsequent sections detail each of the
datasets collected.
3.1.1. Offices
Currently, Hungry operates out of one office location in downtown Boston which was pre-
sented in section 2.1.2 of this study. In addition to Hungry’s existing office location
(71.0507800°W 42.3634400°N), candidate locations for additional offices were also identified.
A zip code shapefile for the Boston downtown area was obtained from the US Census Bureau
and was imported into ArcGIS Pro for processing. The drivetime buffer tool in ArcGIS Pro was
used to create a drive-time polygon using Hungry’s existing office location as the origin point.
Drive-time areas were created for 0-30 minutes away from the origin point. The resulting polygon
area was then used to select zip codes that were within a one-hour drive-time from Hungry’s
existing office location, which can be seen in Figure 4 below. In order to select the most relevant
zip codes, a drive-time buffer was created. The Make Service Area Analysis Layer tool was used
to calculate drive-time away from a facility. Boston’s city center was used as the facility, and a
30-minute cutoff was used. Hungry required their office to be within a 30-minute drive-time from

18
Boston, so zip codes were selected that fell within that the 30-minute drive-time area. Once the
relevant zip codes were selected, the Central Feature tool was used to generate one point-location
in the center of each zip code. The point feature class was exported as the ‘Offices’ feature class
and Hungry’s existing office location was manually added. Existing and Candidate Office Loca-
tions can be seen in Appendix A: Offices.
The resulting ‘Offices’ dataset contains 118 point-locations with five fields, described in
Table 1 below. For each of the zip code point-locations, the Office_Name field was set to the
numerical zip code itself and the Status field was set to ‘candidate.’ Hungry’s existing office
point-location Office_Name field was set to ‘Hungry’ and the Status field was set to ‘existing.’
Figure 4 Office Locations

19

3.1.2. Chefs and Kitchens
Hungry chefs are responsible for sourcing and maintaining their own operations at a li-
censed kitchen. Chefs can cook from leased spaces in commercial kitchens, or they could cook
from their own restaurant. Hungry provided existing chef data including the chef’s name, address,
and other details which can be seen in Table 2 below. The chef data was provided as a .csv format
and this table was imported in ArcGIS Pro as the ‘Chefs’ table. The Geocode Addresses Tool was
used with the default ArcGIS World Geocoding Service and the Chefs table to determine each
chef’s geographic location in coordinates from their address information in the Chefs table. The
Chef’s addresses were geocoded, and the ‘Chefs’ point feature class was created.
Table 1 Offices Feature Class
Table 2 Chefs Feature Class

20
In addition to Hungry’s existing chef’s locations, candidate kitchen locations were also
identified. Hungry provided a .kml file of all commercial kitchens in the United States. The com-
mercial kitchens dataset was imported into ArcGIS Pro. In order to select the most relevant kitch-
ens, a drive-time buffer was created, similar to steps outlined in section 3.1.1. The Make Service
Area Analysis Layer tool was used to calculate drive-time away Boston’s city center, and
30,60,120-minute cutoffs were used. 120-minute cutoff was used as the maximum cutoff time
because Hungry specified that typically, in the Boston market, they would not work with chefs
whose kitchens are over a two-hour drive from Boston. Candidate kitchen locations were selected
that fell within the 30,60, or 120-minute drive-time areas and were exported as the ‘Kitchens’
feature class. The Kitchens feature class field description are shown in Table 3 below.

Next, the Chefs and Kitchens feature classes were then cross-referenced to make sure the
Kitchens feature class had one entry for every Chef listed in the Chefs feature class. If a Chef’s
address was not listed in the Kitchens feature class, it was manually entered into the Kitchens
feature class. Additionally, a new field called ‘KitchenID’ was added to the Chef’s feature class
Table 3 Kitchens Feature Class

21
where each Chef corresponded to one kitchen in the Kitchen feature class. This new field was
updated for each Chef. The Kitchens and Chefs relationship is described in more detail in Section
3.3. The resulting ‘Kitchens’ dataset contains 45 point-locations, shown in Appendix B: Kitchens.
3.1.3. Clients
Hungry’s typical client is a company that has over twenty employees. Clients typically
order catered meals for their staff, and Hungry Captains deliver the food to the clients specified
address. Hungry provided a dataset of existing Clients. The ‘Clients’ table field descriptions can
be seen in Table 4 below.

Similar to section 3.1.2, the Geocode Addresses Tool was used using the default ArcGIS
World Geocoding Service and the Clients table to determine each client’s geographic location.
The resulting ‘Clients’ dataset contains 108 point-locations which represent each client location
and can be seen in Appendix C: Clients.
Table 4 Clients Feature Class

22
3.1.4. Orders
In addition to Hungry’s office, chef’s, and client’s location data, the company also provided
order data from June 1, 2019 to March 31, 2020. There were 788 orders delivered throughout the
Boston market during this time period, which can be seen in Figure 5 below. The green origin
line represents origin kitchen locations, or where the meal was prepared. The line connects each
kitchen location with the corresponding client destination location for each order. Figure 3 shows
that Hungry’s market extends far beyond downtown Boston, with orders originating in Worcester
and Lowell and being delivered to clients in Downtown Boston. This initial representation of
Hungry’s order data confirms that there will be locational inefficiencies in regard to transporta-
tion costs throughout the region.

Figure 5 Hungry's Kitchen to Client Orders

23
The orders data ties all of the previous, existing, data in the datasets presented above to-
gether into one table. The ‘Orders’ table represents all orders Hungry has completed in the Boston
market within the given time period.
Each line in the Orders table represents an order that was placed through the Hungry plat-
form. Each order is given a unique order number and the table records the meal information
including the date it was place, requested to be delivered, who ordered the meal, and how many
people the order is for. The Orders table records what chef the order was placed with as well as
the cost information including how much the client paid, how much the chef receives, and how
much Hungry receives after all necessary account are paid. A detailed description of all the fields
in the Orders table is presented in Table 5 below.

24

Network Analysis
Once the data has been collected and processed, Network Analysis tools are run to create
additional insights. Closest facility and origin destination matrix tools are run to create additional
data that is then added back to the original data tables. The following sections explore these tools
and how the data tables were updated to reflect the new insights that were created.

Table 5 Orders Feature Class

25
3.2.1. Closest Facility
The closest facility tool is used to find the closest facilities to certain locations or ‘inci-
dents.’ First, a closest facilities analysis layer was created where Kitchens were imported as fa-
cilities and Clients were imported as incidents. The closest facilities tool was then run to generate
the 3 closest facilities for each incident using drive-time over ESRI’s road network without taking
traffic into account. The results provide three locations for each of the 108 clients in the Boston
area. Therefore, the resulting ‘routes’ feature class had 324 rows. Each incident (or Client) has
three rows and each row has a field specifying the ClientID and field specifying that ClientID’s
corresponding closest KitchenID. Another field specifies whether the KitchenID is the first, sec-
ond, or third closest facility to its corresponding ClientID. Additionally, there are two other fields
specifying the drive-time miles and time it would take to travel from the ClientID to the Kitch-
enID.
The results were exported as a table and the table was imported into Excel as another tab
in the Clients Table. Excel was then used to generated 9 new fields in the Clients Table, which
can be seen highlighted red in Table 6 below. Using the VLOOKUP function in Excel, each new
field was populated using the corresponding data from the closest facility table that was imported.

26

In addition to updates in the Clients Table, the Kitchens Table was also updated with the
new insights from the closest facility tool, which can be seen highlighted in red in Table 7 below.
Using the PivotTable function in Excel, statistics were calculated for each kitchen. First, a priority
was given to each kitchen based on the number of clients it serves as its closest facility. If a
kitchen serves at least one client as its closest facility, the priority is set to ‘High.’ If a kitchen
serves at least one client as its second closest facility, the priority is set to ‘Medium.’ If a kitchen
serves at least one client as its third closest facility, the priority is set to ‘Low.’ The Priority_Cli-
ents field shows the number of clients that kitchen serves as its designated priority. For example,
if priority = High and priority_clients = 10, then that kitchen is the first closest kitchen to 10
clients. The Total_Clients field shows the number of total clients that kitchen serves as either the
first, second, or third closest facility.
Table 6 Clients

27

3.2.2. Origin Destination Matrices
The origin destination cost matrix tool is used to create transportation costs between a set
of origin and destination locations. First, an origin destination cost matrix layer was created
where the Offices feature class was imported as origins and the Kitchens feature class was im-
ported as destinations. The origin destination tool was then run to generate costs between every
office – kitchen location pair.
There are 118 existing and candidate office locations and 45 existing and candidate
kitchen locations, the resulting origin destination cost matrix has 6,490 different pairs that define
certain transportation costs. The exported table was called the Offices_Kitchens_OD and the
field description are shown in Table 8 below. The Offices_Kitchens_OD describes each OD pair
and presents the transportation costs including minutes and miles it takes to drive from the Of-
fices to the Kitchens.
Table 7 Kitchens
Table 8 Office_Kitchens_OD

28

Similar to the Offices_Kitchens_OD, an origin destination cost matrix layer was created
where the Kitchens feature class was imported as origins and the Clients feature class was im-
ported as destinations. The origin destination tool was then run to generate costs between every
Kitchen – Office location pair.
There are 45 existing and candidate Kitchen locations and 108 existing Client locations,
therefore the resulting origin destination cost matrix has 5,940 different pairs. The exported table
was called Kitchens_Clients_OD and the field descriptions are shown in Table 9 below. The
Kitchens_Clients_OD describes each OD pair and presents the transportation costs including
minutes and miles it takes to drive from the Kitchens to the Clients locations.
Table 9 Kitchens_Clients_OD

29
Database Creation
Now that the data has been collected, processed, and analyzed, the resulting tables need
to be uploaded into a database to develop further insights into Hungry’s operations. First, Mi-
crosoft Access was used to create the database structure by uploading each table and relating
them to each other based on Primary and Foreign Keys. Once the relationships were developed,
the Microsoft Access Database was then uploaded to Microsoft SQL Server where SQL could be
used to query the tables and generate further insights about the data. The insights discovered are
discussed in the next section. An entity relationship model is shown in Figure 6 below.
Figure 6 Database ERD

30
Operations Analysis
Now that the database has been created, we can query the results to find insights about
Hungry’s current operational costs and compare the results to its operations if the company were
to optimize its locations to minimize transportations costs. The results will show how much
money Hungry currently spends on transportation as well as how much money they could save if
they were to optimize its operations.
There are a few assumptions we need to make in order to calculate transportation costs.
The origin destination cost matrix calculated transportation costs incurred for travel across a road
network without taking traffic into consideration. As a result, travel distance (miles) and time
(minutes) were calculated. Hungry provided that it pays its Captains $17 per hour and they drive
vehicles that average 20 miles per gallon. Additionally, they provided that gas is typically $2.50
per gallon. Breaking these numbers down, the study found Hungry’s transportation costs per
mile and per minute. Transportation costs can be found in Table 10 below.

3.4.1. Current Operations
Hungry’s current operation has been defined by the Orders Table, which has information
on the 788 orders it has already conducted in the Boston market. The data shows which kitchen
the meals were cooked from and where the meals were delivered to for each order. The origin-
destination cost matrix gathered insights into transportation costs for transporting supplies from
Table 10 Transportation Costs

31
its office to the kitchens, and the meals from its kitchens to the clients. Using the database created
in section 3.3, and the costs outlined at the beginning of this section, SQL queries were written
to gather insights to determine the transportation costs for Hungry’s current operations. The in-
sights are then used to create a new table called ‘Current_Costs’ that shows the transportation
costs for Hungry’s 788 orders they completed. The Current_Costs Table’s field descriptions can
be seen in Table 11 below.
3.4.2. Optimized Operations
Hungry’s current operations are determined by its clients. The client choses which meals
they would like, and these meals are cooked by a specific chef, at a set location. This reality
introduces inefficiency and waste into the operation where a client could order meals from a chef
that cooks the meals in a location that is far away. In reality, the client’s location could be blocks
away from another kitchen that could be used, saving transportation costs. By optimizing the
location of where the meals are cooked in relation to where the client is located, Hungry could
Table 11 Current Costs

32
save money. By gathering insights into the costs Hungry would save by this optimization, and
comparing them to Hungry’s current operation, this analysis could determine how much money
Hungry could save if it optimized its operations.
Using the database created in section 3.3, and the costs outlined at the beginning of this
section, SQL queries were written to gather insights to determine the transportation costs for
Hungry if it were to use the most optimal locations in terms of optimizing its transportation costs.
Not only do Kitchen-Client location pairs need to be optimize, but an office needs to be
found that optimizing the distance between all kitchen. To determine the best location for a Hun-
gry office, optimal Kitchen-Client location pairs are first assessed to determine the existing or
candidate kitchens that would be used if hungry were to choose the most optimal kitchen for each
client’s order. In this scenario, each client would order from the closest kitchen to the client’s
location. Once the most optimal kitchens are found, an office location is chosen from existing
and candidate office locations that minimizes transportation costs from the office to all of the
optimal kitchens that were selected.
Once the optimal office location has been found, it is used along with the optimal client-
kitchen pairs to calculate transportation costs for each order. Transportation costs are calculated
from the office to the kitchen and then from the kitchen to the client location for each order. These
costs are then used to create a new table called ‘Optimal_Costs’ that shows the transportation
costs for Hungry’s 788 orders if they were to use the most optimal office location and kitchen
locations for each order. The Current_Costs Table’s field descriptions can be seen in Table 12
below.

33

Table 12 Optimized Costs

34
Chapter 4 Results
The methodology used network and operational analysis to create operational insights for
Hungry and recommended actions the company could take to create better efficiencies, and ulti-
mately save money. The results of the analysis also allow Hungry to gather information into its
current operations by looking at historical order data to see how much money they are spending
on transportation costs to deliver meals to their clients. The results also created a comprehensive
database that allows Hungry to answer questions about its current and potentially optimized op-
erations. Ultimately, the result of this study is a dynamic database that can be used to answer any
question that may arise about Hungry’s operations now or in the future. This section will outline
some of the insights that the study set out to answer including how much money Hungry cur-
rently spends on transportation costs and how much money they could save if they optimized
their operations.
From June 1, 2019 to March 31, 2020, Hungry delivered 788 orders in the Boston mar-
ket. Transportation costs were defined as the distance and time it took to drive from Hungry’s
office location to the kitchen where the meals were prepared and then onto the client’s location
where the meals were delivered. It was assumed that Hungry Captains drive vehicles that aver-
age 20 miles per gallon and that gas costs $2.50 on average. Given these costs, it was calculated
that for each mile the Captain drives, it costs Hungry $0.13 in fuel. This accounts only for fuel
costs and no other costs including wear and tear, general service or other unforeseen costs. It was
also assumed that Hungry Captains are paid $17 per hour. At this rate, each minute a Captain
spends in the car costs Hungry $0.28.

35
Current Operations
Using the data that was collected, processed and created throughout the methodology,
this study now looks to analyze the data to find relevant statistics that shed light on Hungry’s
current operations. To put the current operations into perspective, the study pulled summary sta-
tistics from Hungry’s order data which can be seen in Table 13 below.

An average Hungry order costs the customer $853. The average transportation costs for
each order show that Hungry spends 55 minutes traveling 27 miles from their office to the
kitchen to the customer. The summary statistics also show that Hungry Clients order from them
16 times on average, with Hungry’s top customer ordering 112 times within the given time pe-
riod. Breaking the transportation costs down further, the analysis shows that, on average, a Hun-
gry captain spends 27 minutes driving 12 miles from the Hungry office to a kitchen to pick up a
meal, and another 28 minutes driving 15 miles from the kitchens to the client’s location. This
equates to roughly one hour of driving 27 miles for each order, without taking traffic into consid-
eration. Given these results, Hungry spent over $15,000 on transportation costs for the 788 or-
ders during this time period, equating to roughly $20 of transportation costs for every order. The
transportation costs identified here are significant considering Hungry’s average profit per order,
after costs were subtracted to pay the chef, service fee, supplies fee, and taxes, comes out to $39.
Table 13 Order Summary Statistics

36
This means that nearly 50% of Hungry’s profit is taken away in transportation costs alone. With
that being said, the average service fee per order is $14. The average service fee cost does not
fully cover the $20 transportation costs per order, but it does come close. If Hungry raised the
average service fee to $20, they would recoup the majority, if not all, of the transportation costs
incurred.
Optimized Operations
Although Hungry seems to be close to recouping their transportation costs with the $14
average service fee per order, there are still inefficiencies that could be recovered if they decided
to optimize their kitchen locations to deliver the same service at reduced cost.
First, an optimal office location was determined by choosing an office out of the existing
and candidate office locations that minimized transportation costs to all operational kitchens.
Once the optimal kitchens per client were identified, transportation costs were calculated be-
tween each candidate office location and each operational kitchen. The costs were added together
for each candidate office location and the office with the lowest transportation cost was chosen
as the optimal office location. Coincidently, the optimal office location that was identified is
within two miles from Hungry’s current office location as can be seen in Figure 7 below.

37

The next locations to optimize are the kitchens that serve the clients. In section 2.1.3, a
map was shown that displayed Hungry’s current orders and kitchen-client pairs. A similar map
was produced replacing Hungry’s existing kitchens with each client’s optimal kitchen that mini-
mized transportation costs. Figure 8 shows a map of Hungry’s current operations next to another
map of their operations if they were to optimize kitchen locations for delivery their clients. As
you can see on the top map, there are significant distances between the kitchen where the meal is
cooked and the client where the meal was delivered. On the bottom map, the distances have been
significantly reduced and kitchens as the closest kitchens are chosen for each client.

Figure 7 Current vs. Optimized Office Locations

38
If Hungry moved its office and delivered each meal from the client’s optimal kitchen, it
would significantly reduce transportation costs. Section 4.1 identified that Hungry currently
spends roughly $20 per order on transportation costs. By calculating the transportation costs for
the time and distance it took to pick up and deliver the meals for each of the 788 orders, using
Figure 8 Current vs. Optimized Kitchen_Client_OD

39
the optimized locations, it was found that on average a Hungry captain would spend 16 minutes
driving 8 miles from the Hungry office to a kitchen to pick up a meal, and only 6 minutes driving
2 miles from the kitchens to the clients location. This equates to only 22 minutes of driving 10
miles for each order, as opposed to one hour of driving 28 miles in the current situation. Given
these results, Hungry would only spend $7,000 on transportation costs for the 788 orders during
this time period, equating to roughly $8 of transportation costs for every order. By optimizing its
locations, Hungry would cut its transportation costs by nearly 50% compared to its current oper-
ations.

40
Chapter 5 Discussion and Conclusion
Assumptions and Limitations
This study is not meant to be a true model of reality, rather a starting point for future work
that will be discussed in the next section. The model presented in this study makes various as-
sumptions that could be altered to obtain different results. Additionally, this study was limited in
scope and does not cover the full range of criteria that could be implemented to produce results
that better reflect reality. The results and recommendations section of this study, presented in
Chapter 4, reflect the model as it was built. It takes the assumptions into account and was con-
strained by the limitations, all of which will be presented in more depth below.
First, there are many assumptions the model made that need to be taken into consideration.
Many of the assumptions that were made, were decided based on input from Hungry leadership.
Assumptions Hungry provided include Captain cost per hour, average fuel economy for their
vehicles and average fuel price. These assumptions could fluctuate over time and are at the dis-
cretion of Hungry. The assumptions were made under consideration from Hungry leadership and
have been selected to reflect the current reality of Hungry’s operations. The initial assumptions
considered in this initial study could be altered over time to better reflect reality.
Other assumption in this study include relying on ArcGIS Pro’s tools to calculate transpor-
tation costs. Transportation costs were calculated using ESRI’s tools and road transportation net-
work. The tool provided time and distance calculations between and origin and a destination pair.
Transportation between these two points were calculated based on driving times on a Sunday in
July. The tool estimates the time it would take based on these criteria and could be more or less
given the actually travel conditions at the time. Ideally, the travel times presented in this study

41
are the minimum time it would take to travel between two points. Sunday was chosen as the day
to remove rush-hour traffic from consideration and present a baseline travel time where traffic
could later be taken into consideration.
Other assumptions that were made in this study include the logistical aspects of picking up
and dropping off orders. Each meal order is unique and could require more or less effort to fulfill.
It was assumed that one Captain was need for each order, when in reality there could be multiple
captains per order. Additionally, the study does not take additional destinations and time spent at
each location into consideration. Captains could spend a significant amount of time at any one
location picking up and/or dropping off the order. Captains also need to travel back to the Hungry
office or their home at some point. These additional criteria were not taken into consideration.
The operations analysis conducted optimized location pairs based on minimizing transpor-
tation costs in the best-case scenario. In reality, the best-case scenario may not be achievable for
some locations. Just because a client is located next to a candidate commercial kitchen does not
mean that the kitchen could be used. Some kitchen may have high barriers to entry including
lease cost, availability, and other factors that could not be determined until further research was
conducted. However, the best-case scenario that was considered presents a general idea for how
much money could save if it tried to optimize all location pairs.
There are other criteria that should be taken into consideration to present a more realistic
picture of Hungry’s operation, however, the limitations of the study did not allow for more inputs.
The major limitation this study encountered was time. With more time, the limitations presented
below could be implemented into future work.

42
A major limitation encountered in this study was the ability to conduct thorough research
into candidate facility locations. For candidate office locations, zip code areas were used to gen-
eralize point locations instead of researching candidate office locations based on availability and
current market conditions. The candidate office locations and resulting optimal office recommen-
dation was given as a generalized idea of where an office location could go. In future work, more
thought could be considered into the study area demographic and commercial considerations to
determine optimal candidate office facilities. Population density, employment density, and age
demographics are just some of the considerations that should be defined in future work. For this
study, after the recommendation is given, it is assumed that more research could be done to re-
search actual locations to find the best office location. For candidate kitchen locations, the study
relied on a dataset provided by Hungry. There could be additional commercial kitchen within the
market that were not considered in the study and further research could be done to identify addi-
tional candidate commercial kitchen locations. Candidate client locations were not considered in
the study, however, competitive and sales research could be conducted to determine candidate
clients Hungry could target for future business.
Additionally, another limitation the study does not take into consideration is chef transpor-
tation costs. The recommendations outlined above only take Hungry captain transportation costs
into consideration, and it does not take chef transportation costs from traveling from their home
to the commercial kitchen location. If clients were to order meals from their closest commercial
kitchen facility, thereby optimizing the transportation costs, clients could need to sacrifice the
chefs they have the ability to order from as it may not make financial sense for chefs to drive to

43
the client’s closest commercial kitchen location. However, using the results and recommenda-
tions of this study, Hungry could not only assign optimal location pairs, but also limit a client’s
ability to order from certain locations and therefore chefs.
Another limitation that could be considered in future work is weighting locations based on
certain criteria. Not every location should be considered equal as there are many factors above
transportation costs that could be taken into consideration. First, the price of facilities is a large
factor that will fluctuate throughout geographic space. Typically, an office location located down-
town will be more expensive than a location in the suburbs. Similar to office costs, commercial
kitchen’s will also charge different amounts based on location. Clients should also be considered
with different weights based on the value they bring to the company. Some client’s may be repeat
customers that order multiple times a week versus another client that may just order once a month.
Different costs should be considered for all three location types in future work. By weighting each
location based on different costs, it will better reflect reality by weighting their significance com-
pared to other locations within the operation.

Conclusion
Meal delivery is a dynamic market within the food industry that requires geographic con-
siderations to ensure the product is delivered in a timely and cost-effective way. Using geographic
information systems and databases, this study built a model to effectively analyze Hungry Mar-
ketplace’s meal delivery operations in Boston, Massachusetts. The model incorporated existing
operational data as well as candidate facility locations provided by Hungry. The data was pro-
cessed using ArcGIS Pro’s network and spatial analysis tools, which were then incorporated into

44
a database using Microsoft SQL Server. The database was used to gather insights into Hungry’s
operational data to develop results and recommendation that could save Hungry thousands of
dollars in costs. The study found that Hungry could cut its transportation costs in half by opti-
mizing its client and kitchen locations. Additionally, the study recommended optimal candidate
kitchens to use over its existing kitchens in operation. The study recommends that Hungry look
further into its operational data using the database that was created. By answering questions using
the resulting database, Hungry could make changes to its operations in Boston as well as its other
operating location throughout the United States that would result in significant cost savings. The
model presented in this study is not a true representation of reality as there were time constraints
that had to be taken into consideration. However, the study built a solid foundation that could be
built upon in future work to resolve assumptions and include the limitations that were presented
above. Ultimately, the study not only found and recommended optimal locations and operating
guidelines the company could implement to maximize opportunities, but it also produced a com-
prehensive, data-driven, and reproducible methodology for optimizing spatial decisions for com-
panies looking to operate in the meal delivery space within the food industry.

45
REFERENCES
Arafat, Abdulnaser, Iris Patten, and Paul Zwick. 2010. "Site Selection and Suitability Modeling."
2010 ESRI International User Conference Paper #1106.
Blee, Brendan. 2016. "Creating a Geodatabase and Web-GIS Map to Visualize Drone
Legislation in the State of Maryland." A Thesis Presented to the Faculty of the USC
Graduate School University of Southern California In Partial Fulfillment of the
Requirements for the Degree Master of Science (Geographic Information Science and
Technology).
Bosona T, Nordmark I, Gebresenbet G, Ljungberg D. 2013."GIS-Based Analysis of Integrated
Food Distribution Network in Local Food Supply Chain" International journal of
business and management.
Bureau, U.S. Census. 2018. https://censusreporter.org/profiles/31000US14460-boston-
cambridge-newton-ma-nh-metro-area/.
Chopra, Sunil. 2003. "Designing the distribution network in a supply chain." Transportation Re-
search Part E 123-140.
Esri. n.d. "Suitability Analysis." ArcGIS Pro. Accessed 08 08, 2019.
https://pro.arcgis.com/en/pro-app/help/analysis/business-analyst/understanding-suitabil-
ity-analysis.htm.
Giaglis, G. M., Minis, I., Tatarakis, A., & Zeimpekis, V. 2004."Minimizing logistics risk through
real-time vehicle routing and mobile technologies. Research to date and future trends"
nternational Journal of Physical Distribution and Logistics Management 749-764.
Grant, Mason. s.d. "Assessing A Locations Global Connectivity.»

46
Hekmatfar, Farahani &. 2009. "Facility location: concepts, models, algorithms and case studies."
Heidelberg, Germany: Springer-Verlag.
INRIX. 2020. "New Traffic Scorecard Adds Commute Behavior Analysis." Inrix.com. 8 March.
https://inrix.com/blog/2020/03/new-traffic-scorecard-adds-commute-behavior-analysis/.
Jin, Jennifer. 2016. "Installing Public Electric Vehicle Charging Stations: A Site Suitability
Analysis in Los An-geles County, California." University of Southern California.
Jin, Jennifer. 2016. "Installing Public Electric Vehicle Charging Stations: A Site Suitability
Analysis in Los Angeles County, California." USC Thesis.
Lock, S. 2018. Restaurant industry in the U.S. - Statistics & Facts. Accès le 2020.
https://www.statista.com/topics/1135/us-
restaurants/#:~:text=Food%20and%20drinks%20sales%20of,to%20more%20than%2078
2%20billion.
Mahmoudpour, Nina, interviewer par Mason Grant. 2020.
Meixell, Mary, and Vidyaranya Gargeya. 2005. "Global supply chain design: A literature review
and critique." TransportationResearchPartE 531-550.
Melo, M.T., S. Nickel, and F. Saldanha-da-Gama. 2007. "Facility location and supply chain
management – A review." European Journal of Operational Research 401-412.
Mitchell, Andy. 2012. "The Esri Guide to GIS Analysis Volume 4: Modeling Suitability,
Movement, and Interaction." ESRI Press.
Oliver, David. 2015. Why a new food tech wave is shaking up the industry.
https://www.grocerydive.com/news/why-a-new-food-tech-wave-is-shaking-up-the-
industry/535772/.

47
WorldAtlas. 2019. Which US City Has The "Most Innovative Square Mile On The Planet"? 24
January. https://www.worldatlas.com/articles/which-us-city-has-the-most-innovative-
square-mile-on-the-planet.html.

48
APPENDIX A: Offices

49
APPENDIX B: Kitchens

50
APPENDIX C: Clients

Abstract (if available)

Abstract The food industry has been completely disrupted over the past 5 years with the onset of platforms like Doordash, Instacart, HelloFresh, and Hungry Marketplace. These platforms, and others like it, offer customers timely, cost-saving, and convenient ways to prepare, consume, and/or experience meals. The explosion in food-industry innovation has changed the dynamic of food services altogether as status quo operations aren't meeting the needs of the innovative operations. More people are opting in to having their meals delivered to more comfortable, familiar, locations as opposed to going to brick-and-mortar restaurants to consume their meals. As a result, the food industry’s changing dynamic has led many to reconsider owning expensive restaurant locations in consideration of alternative commercial locations that are much cheaper and allow them to deliver the same quality of service. Given the importance of kitchen and office locations and the critically changing nature of the industry, new methods are needed to determine the optimal locations for companies that are delivering services in this new environment. Using a GIS (Geographic Information System), this thesis uses spatial analysis including site-suitability analysis and network analysis to build an optimization model for Hungry Marketplace, a food-industry start-up. The model utilized the company’s current operations in Boston, Massachusetts as a case study. The model identified optimal locations for kitchen and warehouse operations that maximize the market opportunities while keeping the operational expenses low. This thesis provides recommendations to the company for a cost-effective operation going forward. Ultimately, this data-driven and reproducible methodology can be applied by existing and potential companies in the food industry for optimizing their spatial decisions.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Enriching the Demographic Survey sampling for the Los Angeles County Annual Homeless Count with spatial statistics

PDF

Exploring remote sensing and geographic information systems technologies to understand vegetation changes in response to land management practices at Finke Gorge National Park, Australia Between ...

PDF

Using the Digital Shoreline Analysis System (DSAS) to analyze changes in shoreline position caused by seawalls along a section of Oregon's coast

PDF

Impacts of vegetation management on wildfire severity: a study of the 2021 Caldor fire

PDF

Using GIS to perform a risk assessment for air-transmitted bioterrorism within San Diego County

PDF

Geographic information systems and marketing: a transdisciplinary approach to curriculum development

PDF

Building a spatial database for agricultural record keeping and management on a regenerative farm

PDF

The use of site suitability analysis to model changes in beach geomorphology due to coastal structures

PDF

Generating bicyclist counts using volunteered and professional geographic information through a mobile application

PDF

Creating a Web GIS to support field operations and enhance data collection for the Animal and Plant Health Inspection Service (APHIS)

PDF

Bringing GIS to a small community water system

PDF

Assessing modern conflict to monitor human rights with remote sensing: Russia's war in Ukraine

PDF

Applying GIS to landscape irrigation systems: a case study of the Music Academy of the West campus in Montecito, CA

PDF

The impact of severe coastal flooding on economic recovery disparities: a study of New Jersey communities following Hurricane Sandy

PDF

Surface rupture detection with support vector machine classification: case study from Ridgecrest, CA

PDF

Geodatabase for archaeogenetics: ancient peoples and family lines

PDF

Classifying ash cloud attributes of Eyjafjallajökull volcano, Iceland, using satellite remote sensing

PDF

A threat-based least-cost path decision support model for national security resource allocation along the US-Mexico border

PDF

Development and evaluation of a USV based mapping system for remote sensing of eelgrass extent in southern California

PDF

Scenario-based site suitability analysis and framework for biodiversity conservation: agricultural zone, Galapagos Archipelago, Ecuador

Asset Metadata

Creator Grant, Mason (author)

Core Title Operational optimization model for Hungry Marketplace using geographic information systems

School College of Letters, Arts and Sciences

Degree Master of Science

Degree Program Geographic Information Science and Technology

Degree Conferral Date 2021-08

Publication Date 08/04/2021

Defense Date 08/27/2020

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag geographic information systems,OAI-PMH Harvest,operations,optimization

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Fleming, Steven (committee chair), Ruddell, Darren (committee member), Wu, An-Min (committee member)

Creator Email masonbgrant@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC15676562

Unique identifier UC15676562

Legacy Identifier etd-GrantMason-9985

Document Type Thesis

Format application/pdf (imt)

Rights Grant, Mason

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu