DCF Week 5 Warm-Up for Assignment

Do this work in an Rmd file named Week-5-Warmup-XXX.Rmd.¹ Rather than typing commands at the console, type them into a chunk and run that chunk in the console. (If you’re not sure what this means, ask! There is a keyboard shortcut that makes it easy.) When the chunk does what you want, compile the Rmd document to HTML. Then move on to the next task and repeat the cycle: compose, get it working, compile to HTML.

You will use four data tables in this exercise:

1. ZipGeography in the DCF package.
2. Restaurants which you must load into R.
3. Cuisines …………. ditto …
4. ViolationCodes ………ditto …

Read in (2), (3), and (4) with these commands:

load( url( "http://tinyurl.com/m4o4n2b/DCF/ViolationCodes.rda" ) ) 
load( url( "http://tinyurl.com/m4o4n2b/DCF/Cuisines.rda" ) )
load( url( "http://tinyurl.com/m4o4n2b/DCF/Restaurants.rda" ) )

Do this now, before reading on. It will take about 2 minutes for the last one. Then, while you’re waiting, read the rest of this activity. You’ll know it’s working if Restaurants, Cuisines, ViolationCodes show up in your account.

How’s the Food?

Government agencies have increasingly been putting data in publicly accessible places. For example, in India, the http://attendance.gov.in/ website tracks the attendance at work of government employees.²

New York City publishes many datasets, including health inspections of restaurants. That’s what you’re going to work on now.

The data table Restaurants contains information about each health violation. (See the introduction for how to access the data table.) Note that the DBA variable contains the name of the restaurant.³

• What is the meaning of a case in this data table? You can look at the data table with this command: View( Restaurants ).

• How many cases are there?

• How many distinct restaurant names are there?

• Are there any restaurants with the same name but with multiple branch locations? Select one or more variables that plausibly identify a unique branch.

• Which restaurants (individual branches) have the most violations?

The data table ViolationCodes contains a description of the different types of violations and whether they are critical. Cuisines gives the meaning for the CUISINECODE variable.

• How many critical violations were reported? Non-critical?

• Which restaurants have the greatest number of “critical” violations? Produce a table showing the number of critical and non-critical violations for each restaurant.

• What’s the most common cuisine type in the whole city? In each borough? In each zip code?

The data table Cuisines details the code for each restaurant’s cuisine type.

• What kind of cuisine has the most violations per restaurant, on average?

You can create a table where the case is “an individual restaurant branch” with this statement:

RR <- Restaurants %>% 
  group_by( PHONE ) %>% 
  filter( row_number(PHONE)==1 )

Get the latitude and longitude of each zip code from ZipGeography. Plot out the location of each restaurant, using the borough for color. Use the aesthetic

position=position_jitter( width=0.02, height=0.02 )

to spread out the restaurants a bit. Play with alpha= to get a nice graphic. Which borough corresponds to each borough number?

• Choose 3 or 4 cuisines of interest to you. Plot out the location of each restaurant of that type as an additional layer on the previous graphic.

• Find the violations that are most likely to be repeated for a given restaurant branch. (If a violation appears twice or more for a given restaurant branch, the violation extended over more than one inspection period.)

• Make an informative plot of the distribution of SCORES for each letter CURRENTGRADE.

• Do the same for the distribution of SCORES by ACTION.

---

1. XXX should be replaced by your personal ID, e.g. your initials.↩

2. See this New York Times article.↩

3. The data tables are published by the New York City government as a zip file at https://data.cityofnewyork.us/Health/Restaurant-Inspection-Results/4vkw-7nck?↩