Project Team 3: Philly Food Delivery, DS4200 F20

Ming Yuen Thomas Cheong, Benjamin Jacobson, Christopher Sedayao

Service-Learning Course Project as part of DS 4200 F20: Information Visualization, taught by Prof. Cody Dunne, Data Visualization @ Khoury, Northeastern University.

Abstract

Most mature delivery restaurants lose money on their delivery orders. This is because of the premiums introduced by do-it-all delivery services like GrubHub or DoorDash. This project aims to leverage data that we get from our undisclosed partner in the logistics industry to provide insight and analysis on delivery cost efficiency. The goal is to enable restaurants to make better informed operational decisions that will be beneficial for maintaining profit margins in the delivery market. Restaurant owners do not necessarily have the time to dig into nuanced financial information, so our project is an attempt to provide insightful content at-a-glance. Based on our dataset of the Philadelphia delivery market, most restaurants have the ability to save significantly by outsourcing delivery to delivery-only logistics companies.

Visualization

• The 10 restaurants represent the top 10 restaurants with the highest order volume in 2020.
• Delivery zones are delivery radiuses that our partner operates within. These delivery zones have a maximum delivery time of 10 minutes which allow our partner company to operate efficiently.
• Cost savings represent the 20% premium restaurants pay for competitors. This is what restaurants can save if they convert to our partner company. In order to make our findings realistic, our cost saving calculation assumes that restuarants can convert 50% of competitor orders within these delivery zones to our partner.
• Hover over restaurant bar charts to reveal specific restaurant data.
• When not hovering over the bar chart, pie chart and cost savings data reveal information for all 10 restaurants.

Demo Video

Visualization explanation

The bar charts help visualize the top 10 restaurants and their related order volume data as well as their average distance (in miles) for all orders. We added hover-linking through both bar charts that will change the color of the hovered bar to show the relation between the two charts. There is also tool-tip included for the user to find the specific order volume or average order distance for that specific restaurant. Given that information, we implemented details on demand pie charts for the bar the user ends up hovering over. The pie charts help distinguish the distribution of order origin (grubhub, snackpass, etc...) as well as the distribution of orders outside delivery zones for that specific restaurant. If no bars are hovered, then the pie charts represent data for all 10 restaurants, whereas, if a bar is hovered, only that specific restaurant data is shown. To elevate our project further, we decided to include a cost reduction aspect to our visualizations. The cost reduction/saving represents what restaurants can save if they convert competitor orders to our partner company. Similar to the pie chart, the cost savings is a detail on demand where if no restaurant bar is hovered, it represents total cost savings by all 10 restaurants.

Link to Presentation

Acknowledgments

List here where any code, packages/libraries, text, images, designs, etc. that you leverage come from.

• D3: Data-Driven Documents by Mike Bostock is used for manipulating the DOM to create visualizations.
• Pure CSS responsive 'Fork me on GitHub' ribbon by Chris Heilmann is used to create the banner that links back to the source code repository.
• Count fields from csv dataset in d3 js from Stackoverflow was used to create a map of each restaurant and a distinct count of a variable.
• d3 grouping from observablehq.com was used to learn how to group data using d3 and how to use rollup functions.
• d3 bar chart by TutorialsTeacher was used to learn how to make ordinal bar charts.
• Sort array numerically by Derek Austin was used to figure out how to sort an array numerically.
• Pie Charts by Yan Holtz was used to figure out how to create a pie chart in D3.
• D3 Chart Legends by Yan Holtz was used to figure out how to create a chart legend in D3.
• Color Brewer was used to figure choose colors for pie charts.
• W3Schools was used to learn about tooltip styles.
• D3 tooltips tips by Chris Viau was used to learn how to use tooltips.
• D3 format comma by Michael Stanaland was used to learn how to format commas.