Eat at State

A common theme in the lab is to ask one another what do you have for lunch today? Sometimes we don’t bring anything and our options are to go to the international center food court, the restaurants on Grand River or the residential dining halls. MSU employees have a great option to buy admission into the dining hall at a discounted rate so I always tag along with them so they can swipe me in. We typically preview the menus at eatatstate.msu.edu and will choose to go to the dining hall one with the best menu. One of my favorite dishes from the cafeteria is Presidential Chicken which is a chicken dish with an alfredo like sauce with vegetables and is wrapped up in puff pastry envelope. It tastes great and is exponentially better than the food I’ve had at weddings. We haven’t seen the dish on the menu in well over a year so I decided to see if I could scrape the menus and make a quick shiny dashboard to figure out the next time they would serve it.

library(rvest)
library(knitr)
library(kableExtra)
library(tidyverse)
library(lubridate)

Generate urls

The key to generating a list of URLs for scraping is to identify the patterns of interest and stick them together using the paste() function. Here I used the main eatatstate domain and simply created an object for each dining hall. Then I can use the paste function to link them together so then I have URLs for each dining hall. Then I generated a sequence of dates and appended it to the dining hall URLs.

domain <- "https://eatatstate.msu.edu/menu/"
shaw <- "The%20Vista%20at%20Shaw/all/"
sny_phy <- "The%20Gallery%20at%20Snyder%20Phillips/all/"
landon <- "Heritage%20Commons%20at%20Landon/all/"

dining_halls <- c(shaw, sny_phy, landon)

# List of dates
future_date <- ymd(today()) + days(10)
dates <- seq(ymd(today()), future_date, by = 'day')

# Number of dining halls
dining_hall_no <- length(dining_halls) %>%
  unlist()

# Create lunch urls
day_url <- dining_halls %>% 
  map(~paste(domain, .x, dates, sep = "")) %>%
  unlist()

dining_halls

## [1] "The%20Vista%20at%20Shaw/all/"               
## [2] "The%20Gallery%20at%20Snyder%20Phillips/all/"
## [3] "Heritage%20Commons%20at%20Landon/all/"

dates

##  [1] "2019-12-19" "2019-12-20" "2019-12-21" "2019-12-22" "2019-12-23"
##  [6] "2019-12-24" "2019-12-25" "2019-12-26" "2019-12-27" "2019-12-28"
## [11] "2019-12-29"

day_url %>% 
  head()

## [1] "https://eatatstate.msu.edu/menu/The%20Vista%20at%20Shaw/all/2019-12-19"
## [2] "https://eatatstate.msu.edu/menu/The%20Vista%20at%20Shaw/all/2019-12-20"
## [3] "https://eatatstate.msu.edu/menu/The%20Vista%20at%20Shaw/all/2019-12-21"
## [4] "https://eatatstate.msu.edu/menu/The%20Vista%20at%20Shaw/all/2019-12-22"
## [5] "https://eatatstate.msu.edu/menu/The%20Vista%20at%20Shaw/all/2019-12-23"
## [6] "https://eatatstate.msu.edu/menu/The%20Vista%20at%20Shaw/all/2019-12-24"

Next we will generate the URLs for lunch and dinner and combine them all.

# Create dinner urls
dinner <- "?field_mealtime_target_id=191"

night_url <- dining_halls %>% 
  map(~paste(domain, .x, dates, dinner, sep = "")) %>%
  unlist()

# Combine url list
all_url <- c(day_url, night_url)

all_url %>% 
  head()

## [1] "https://eatatstate.msu.edu/menu/The%20Vista%20at%20Shaw/all/2019-12-19"
## [2] "https://eatatstate.msu.edu/menu/The%20Vista%20at%20Shaw/all/2019-12-20"
## [3] "https://eatatstate.msu.edu/menu/The%20Vista%20at%20Shaw/all/2019-12-21"
## [4] "https://eatatstate.msu.edu/menu/The%20Vista%20at%20Shaw/all/2019-12-22"
## [5] "https://eatatstate.msu.edu/menu/The%20Vista%20at%20Shaw/all/2019-12-23"
## [6] "https://eatatstate.msu.edu/menu/The%20Vista%20at%20Shaw/all/2019-12-24"

# Date
date <- rep(dates, times = 2 * dining_hall_no)

# Cafe

no_dates <- length(dates) 
names <- str_replace_all(dining_halls, "%20|\\/all\\/"," ")
cafe <- rep(names, each = no_dates, times = 2)

# Time
time <- rep(c("Lunch", "Dinner"), each = no_dates * dining_hall_no)
time

##  [1] "Lunch"  "Lunch"  "Lunch"  "Lunch"  "Lunch"  "Lunch"  "Lunch"  "Lunch" 
##  [9] "Lunch"  "Lunch"  "Lunch"  "Lunch"  "Lunch"  "Lunch"  "Lunch"  "Lunch" 
## [17] "Lunch"  "Lunch"  "Lunch"  "Lunch"  "Lunch"  "Lunch"  "Lunch"  "Lunch" 
## [25] "Lunch"  "Lunch"  "Lunch"  "Lunch"  "Lunch"  "Lunch"  "Lunch"  "Lunch" 
## [33] "Lunch"  "Dinner" "Dinner" "Dinner" "Dinner" "Dinner" "Dinner" "Dinner"
## [41] "Dinner" "Dinner" "Dinner" "Dinner" "Dinner" "Dinner" "Dinner" "Dinner"
## [49] "Dinner" "Dinner" "Dinner" "Dinner" "Dinner" "Dinner" "Dinner" "Dinner"
## [57] "Dinner" "Dinner" "Dinner" "Dinner" "Dinner" "Dinner" "Dinner" "Dinner"
## [65] "Dinner" "Dinner"

# Menu 
menu <- all_url %>%   
 map(~read_html(.x) %>% 
      html_nodes(".meal-title") %>% 
      html_text()) 

food <- tibble(date, cafe, time, menu) %>% 
  unnest(menu) 

dim(food)
food

## [1] 2349    4

food %>% 
  filter(str_detect(menu, "Presidential"))

## # A tibble: 1 x 4
##   date       cafe                       time   menu                
##   <date>     <chr>                      <chr>  <chr>               
## 1 2018-10-05 Heritage Commons at Landon Dinner Presidential Chicken

food %>% 
  filter(date == today())

## # A tibble: 0 x 4
## # … with 4 variables: date <date>, cafe <chr>, time <chr>, menu <chr>

Wordcloud

Since we have a tidy data frame of all the dining halls we can use the tidytext package from Julia Silge and David Robinson to tokenize the menu and make a word cloud. First I used the palette from MSU’s website and set it as an object pal. Next we will unnest the tokens from the food data frame and count up the most frequent words. We can remove filler words with an anti_join() command and then tell R to create a wordcloud!

library(tidytext)
library(wordcloud)
library(RColorBrewer)

# MSU palette colors  
pal <- c("#97A2A2","#F08521","#6E005F","#F08521","#97A2A2","#0DB14B")

set.seed(1234)

food %>% 
  unnest_tokens(menu, menu) %>% 
  count(menu, sort = TRUE) %>% 
  rename(word = menu) %>% 
  anti_join(stop_words, by = "word") %>% 
  with(wordcloud(word, n, random.order = FALSE, max.words = 40, color = pal,fixed.asp = TRUE))

Shiny Dashboard

Now that we learned how to make a wordcloud, scrape the menu into a tidy data frame we can work on making a interact with the data by making a shiny dashboard.

library(shiny)
library(shinydashboard)

ui <- dashboardPage(skin = "green",
  dashboardHeader(title = "Eat at State"),
  dashboardSidebar( selectInput(inputId = 'dininghall',
                  label = 'Choose Dining Hall:',
                  choices = c("All","The Vista at Shaw ",
                              "The Gallery at Snyder Phillips ",
                              "Heritage Commons at Landon "))
  ),
  
  dashboardBody(
    # Boxes need to be put in a row (or column)
    fluidRow(
      box(dataTableOutput("menu"),title = "MSU Dining Menu",
          width = 12, status = "primary")
      )
    )
  )

server <- function(input, output) {
  
   menu <- reactive({
    if(input$dininghall == "All")
        return(food)
    else
      food %>% 
       filter(cafe %in% input$dininghall)
  })

  output$menu <- renderDataTable(menu())
}

shinyApp(ui, server)

Eat at State Shiny Dashboard

Presidential Chicken!

I had the idea to scrape for presidential chicken on Monday this week and I was thrilled that Landon had it for dinner on Friday. Now that I was able to go I decided to create this blog post to share with everyone!