Las Vegas Restaurant Grades
Data Analysis
Project Goal
I downloaded the City of Las Vegas restaurant inspection grades dataset from the city's open data library to explore trends and insights through exploratory data analysis. With ten years' worth of inspection grades, I was curious to see if patterns would emerge, such as shifts in restaurant performance over time, common health code violations, or differences in grades based on location or type of restaurant. This dataset offered an opportunity to uncover valuable insights into the health and safety standards of Las Vegas restaurants across a decade.
Cleaning and Exploratory Data Analysis
You can look at the actual notebook to see the code I used to enact each of the steps I describe below. It’s messy because that’s how cleaning and exploratory data analysis go. Knowledge emerges as you dig further.
Click here to view the colab notebook.
A big part of exploratory data analysis (EDA) is cleaning and reshaping the data so that it becomes easier and more intuitive to work with when we get to the statistical analysis phase. Following are the steps I took during the cleaning process with my thoughts:
df = rest.copy()
1) Of course, I import the data. Then I make a copy of it to leave the raw file intact so that I can refer to it again if necessary.
df.head()
2) I take a quick look at my data. This automatically lets me see the columns and first five rows of my dataframe.
# Split Location_1 column by latitude and longitude
df[['Latitude', 'Longitude']] = df['Location_1'].str.split(',', expand=True)I see immediately that I want to split up my Location_1 column into longitude and latitude. I know I will use these later, so I might as well do it now!
I’ve only seen letter grades of A, B, and C in person when dining out, so I want a list of possible grades with their count.
Inspection_Grade Inspection_Result Count
B Follow Up Required 1
X No Further Action 1
B Downgrade 1
A Grade 1
P Compliance Schedule 1
F Non-compliant 1
C Unspecified 1
Closed with Fees 1
A Grade 1
B Closed with Fees 1
A PR Change of Ownership Required 1
a Inactive Status 1
A Complaint Invalid/Unsubstantiated 1
Complaint valid with fee 1
Complaint valid without fee 1
Case Transfer/Referral 1
Closed without Fees 1
C Downgrade 1
S Permit Suspended 2
B C Downgrade 2
A Unspecified 3
O Approved - Follow Up: Operations 3
C B Downgrade 3
X Permit Suspended 4
A Compliance Schedule 4
N Not Approved 4
A B Downgrade 14
Follow Up Required 16
B A Grade 37
A Inactive Status 49
O Approved 52
A No Further Action 148
X Closed without Fees 175
Closed with Fees 1,402
C C Downgrade 3,608
P Compliant 7,537
B B Downgrade 16,735
A Compliant 42,844
A Grade 126,663
Consolidate
Though these grades are interesting, most are too few, and some are blank, so they aren’t useful for data analysis. I will consolidate them into a handful of aggregate grades or eliminate them from the analysis.
I want something that adheres to the existing grading scale but is easily understandable at first glance, like the school grade system: A for best and F for failure, so A, B, C, D, and F.
My thinking for the grade buckets:
A grade: the restaurant is doing great!
B grade: the restaurant is fine but can do better
C grade: meets the minimum compliance standards
D grade: struggling, but still open
F grade: closed
For the rest, I discarded the miscellaneous single counts.