VenueSignal

Overview

The hospitality and restaurant industries face a critical challenge when opening new locations: understanding how operational constraints, such as limited parking, will affect customer satisfaction and reviews. For businesses considering expansion into urban city centers where parking is scarce, there's a significant financial risk, as customer reception is hard to predict. This is especially relevant in existing cities with large surface parking lots where debate over ideal development practices regularly swirls, seeking to weigh the needs of the local economy, business owners, and consumers.

This project aims to develop a machine learning model to predict Yelp review ratings and sentiment for new businesses opening in city centers with limited parking. By leveraging the Yelp Open Dataset, which contains millions of reviews across various business categories and locations, we can identify patterns that correlate parking constraints with customer feedback.

The model's objective is to provide actionable insights for business owners and investors by forecasting expected review scores and identifying key factors that influence customer satisfaction in parking-constrained environments. This is fundamentally a supervised learning problem—specifically, a regression task if we're predicting star ratings (1-5 scale) or a classification task if we're categorizing sentiment (positive/negative/neutral). Additionally, we may employ natural language processing (NLP) techniques to extract sentiment and themes from review text. Understanding these dynamics will help stakeholders make informed decisions about location selection, pricing strategies, and operational adjustments to mitigate parking-related concerns.

Data Source

Our data source is the Yelp Open Dataset, sourced directly from Yelp here: https://business.yelp.com/data/resources/open-dataset/

Per the Yelp terms:

The Data is made available by Yelp Inc. (“Yelp”) to enable you to access valuable local information to develop an academic project as part of an ongoing course of study or for non-commercial purposes.

The dataset consists of five JSON files:

1. yelp_academic_dataset_business.json

• Records: 150,346 businesses
• Size: 113 MB
• Contains: Business information including name, location, categories, ratings, hours, and attributes
• Sample fields: business_id, name, address, city, state, postal_code, latitude, longitude, stars, review_count, is_open, attributes, categories, hours

2. yelp_academic_dataset_checkin.json

• Records: 131,930 check-ins
• Size: 274 MB
• Contains: Check-in timestamps for businesses
• Sample fields: business_id, date (comma-separated list of check-in timestamps)

3. yelp_academic_dataset_review.json

• Records: 6,990,280 reviews (largest dataset)
• Size: 5.0 GB
• Contains: User reviews with ratings and text
• Sample fields: review_id, user_id, business_id, stars, useful, funny, cool, text, date

4. yelp_academic_dataset_tip.json

• Records: 908,915 tips
• Size: 172 MB
• Contains: Short tips/suggestions from users
• Sample fields: user_id, business_id, text, date, compliment_count

5. yelp_academic_dataset_user.json

• Records: 1,987,897 users (second largest)
• Size: 3.1 GB
• Contains: User profiles with social network info and compliments
• Sample fields: user_id, name, review_count, yelping_since, friends (list), useful, funny, cool, fans, average_stars, compliment_hot/more/profile/etc.

License Terms

This project is part of the MSAAI program at the University of San Diego.

S3 Bucket

The JSON files for this project (as described above) have been loaded into an S3 bucket for processing. The bucket is located at s3://yelp-aai540-group6/yelp-dataset/.

Additionally, the ARN for the bucket is arn:aws:s3:::yelp-aai540-group6.

Permissions for the bucket have been made public via AWS Learner Labs.

Table Documentation

yelp.business

Field Name	Data Type	Description
business_id	string	Unique Yelp identifier for the business
name	string	Business name
address	string	Street address
city	string	City where the business is located
state	string	State or province code
postal_code	string	ZIP or postal code
latitude	double	Latitude coordinate
longitude	double	Longitude coordinate
stars	double	Average Yelp star rating
review_count	int	Total number of reviews
is_open	int	Business open status (1 = open, 0 = closed)
attributes	map<string,string>	Raw semi-structured business attributes. Utilize yelp.business_attributes
categories	string	Comma-separated business categories
hours	map<string,string>	Mapping of day → opening hours. Utilize yelp.business_attributes

yelp.review

Field Name	Data Type	Description
review_id	string	Unique Yelp identifier for the review
user_id	string	Identifier of the user who wrote the review
business_id	string	Identifier of the reviewed business
stars	double	Star rating given in the review
useful	int	Number of “useful” votes
funny	int	Number of “funny” votes
cool	int	Number of “cool” votes
text	string	Full review text
date	string	Date the review was posted

yelp.user

Field Name	Data Type	Description
user_id	string	Unique Yelp identifier for the user
name	string	User display name
review_count	int	Total number of reviews written
yelping_since	string	Date the user joined Yelp
friends	array	List of Yelp user IDs representing friends
useful	int	Total “useful” votes received
funny	int	Total “funny” votes received
cool	int	Total “cool” votes received
fans	int	Number of fans
elite	array	Years the user was an Elite member
average_stars	double	Average rating given by the user
compliment_hot	int	Count of “hot” compliments
compliment_more	int	Count of “more” compliments
compliment_profile	int	Profile compliments
compliment_cute	int	Cute compliments
compliment_list	int	List compliments
compliment_note	int	Note compliments
compliment_plain	int	Plain compliments
compliment_cool	int	Cool compliments
compliment_funny	int	Funny compliments
compliment_writer	int	Writer compliments
compliment_photos	int	Photo compliments

yelp.checkin

Field Name	Data Type	Description
business_id	string	Identifier of the business being checked into
date	string	Comma-separated list of check-in timestamps

yelp.tip

Field Name	Data Type	Description
user_id	string	Identifier of the user who wrote the tip
business_id	string	Identifier of the business
text	string	Tip text
date	string	Date the tip was posted
compliment_count	int	Number of compliments received

yelp.business_attributes

Field Name	Data Type	Description
business_id	string	Unique Yelp business identifier
acceptsinsurance	string	Indicates whether insurance is accepted
agesallowed	string	Age restrictions for the business
alcohol	string	Alcohol service type
bikeparking	string	Bike parking availability
businessacceptsbitcoin	string	Bitcoin acceptance indicator
businessacceptscreditcards	string	Credit card acceptance indicator
byappointmentonly	string	Appointment-only business
byob	string	Bring-your-own-beverage allowed
byobcorkage	string	Corkage policy
caters	string	Catering availability
coatcheck	string	Coat check availability
corkage	string	Corkage details
dogsallowed	string	Dogs allowed
drivethru	string	Drive-thru availability
goodfordancing	string	Suitable for dancing
goodforkids	string	Suitable for kids
happyhour	string	Happy hour availability
hastv	string	TV availability
music	string	Music options
noiselevel	string	Noise level category
open24hours	string	Open 24 hours
outdoorseating	string	Outdoor seating availability
restaurantsattire	string	Dress code
restaurantscounterservice	string	Counter service availability
restaurantsdelivery	string	Delivery availability
restaurantsgoodforgroups	string	Suitable for groups
restaurantspricerange2	string	Price range (1–4)
restaurantsreservations	string	Reservations accepted
restaurantstableservice	string	Table service availability
restaurantstakeout	string	Takeout availability
smoking	string	Smoking policy
wheelchairaccessible	string	Wheelchair accessibility
wifi	string	WiFi availability/type
parking_garage	boolean	Garage parking available
parking_street	boolean	Street parking available
parking_validated	boolean	Validated parking available
parking_lot	boolean	Parking lot available
parking_valet	boolean	Valet parking available
ambience_divey	boolean	Divey atmosphere
ambience_hipster	boolean	Hipster atmosphere
ambience_casual	boolean	Casual atmosphere
ambience_touristy	boolean	Touristy atmosphere
ambience_trendy	boolean	Trendy atmosphere
ambience_intimate	boolean	Intimate atmosphere
ambience_romantic	boolean	Romantic atmosphere
ambience_classy	boolean	Classy atmosphere
ambience_upscale	boolean	Upscale atmosphere
good_for_dessert	boolean	Good for dessert
good_for_latenight	boolean	Good for late night
good_for_lunch	boolean	Good for lunch
good_for_dinner	boolean	Good for dinner
good_for_brunch	boolean	Good for brunch
good_for_breakfast	boolean	Good for breakfast
bestnight_monday	boolean	Best night: Monday
bestnight_tuesday	boolean	Best night: Tuesday
bestnight_wednesday	boolean	Best night: Wednesday
bestnight_thursday	boolean	Best night: Thursday
bestnight_friday	boolean	Best night: Friday
bestnight_saturday	boolean	Best night: Saturday
bestnight_sunday	boolean	Best night: Sunday
hair_africanamerican	boolean	African-American hair specialization
hair_asian	boolean	Asian hair specialization
hair_coloring	boolean	Hair coloring services
hair_curly	boolean	Curly hair services
hair_extensions	boolean	Hair extensions services
hair_kids	boolean	Kids hair services
hair_perms	boolean	Perm services
hair_straightperms	boolean	Straight perm services
dairy_free	boolean	Dairy-free options
gluten_free	boolean	Gluten-free options
vegan	boolean	Vegan options
kosher	boolean	Kosher options
halal	boolean	Halal options
soy_free	boolean	Soy-free options
vegetarian	boolean	Vegetarian options
hours_monday	string	Monday operating hours
hours_tuesday	string	Tuesday operating hours
hours_wednesday	string	Wednesday operating hours
hours_thursday	string	Thursday operating hours
hours_friday	string	Friday operating hours
hours_saturday	string	Saturday operating hours
hours_sunday	string	Sunday operating hours
open_days_count	int	Number of days with defined hours
open_on_weekend	boolean	Indicates weekend availability

Order of Notebooks in this Project.

1. EDA
2. AthenaTables
3. DataLake
4. FeatureStore
5. BENCHMARK
6. XGBoost
7. ModelDeployment
8. ModelComparison
9. CI/CD

Contributors

Carrie Little

Dean P. Simmer

Omar Sagoo