Introduction

In February 4, 2024, I completed the R Programming course which proved to be quite a challenge as a beginner at the time. Five months later, I wanted to give it another try to simplify the functions required to pass for other learners that may have encountered complicated solutions online.

Required files : Dataset - Dataset Documentation - Assignment

Preparing Data

# Used Libraries
library(dplyr)

# Loading the dataset
data <- read.csv("outcome-of-care-measures.csv", colClasses = "character")

# Extracting and cleaning useful columns
hospitals <- data |>
  mutate(
    heart_attack = Hospital.30.Day.Death..Mortality..Rates.from.Heart.Attack,
    heart_failure = Hospital.30.Day.Death..Mortality..Rates.from.Heart.Failure,
    pneumonia = Hospital.30.Day.Death..Mortality..Rates.from.Pneumonia,
    State,
    hospital_name = Hospital.Name,
    .keep = "none"
  )

# Converting the values back to numeric and handling the `Not Available` cases
outcomes <- c("heart_attack", "heart_failure", "pneumonia")

hospitals[outcomes] <- lapply(hospitals[outcomes], function(x) {
  as.numeric(ifelse(x == "Not Available", NA, x))
})

Finding the best hospital in a state

Code

best <- function(state, outcome) {
  if (!(state %in% hospitals$State)) {
    return("invalid state") # Replace with stop() to comply with the assignment conditions
  }
  
  outcome <- tolower(outcome)
  outcome_col <- switch(outcome,
                        "heart attack" = "heart_attack",
                        "heart failure" = "heart_failure",
                        "pneumonia" = "pneumonia",
                        return("invalid outcome")) # Replace with stop() to comply with the assignment conditions

  hospitals |>
    arrange(hospital_name) |>
    filter(State == state) |>
    filter(!is.na(!!sym(outcome_col))) |>
    slice_min(!!sym(outcome_col), with_ties = FALSE) |> 
    pull(hospital_name)
  
}

I replaced stop() with return() in order to prevent it from stopping the rendering of my Quarto document.

After checking whether the state abbreviation inserted by the user is found or not, a best practice is converting the demanded outcome to lowercase.

filter(!is.na(!!sym(outcome_col))) is used to filter any NA values in the specified outcome column using sym() that converts a string into a symbol which then gets unquoted by !!, for more information check the documentation by clicking on sym().

Testing

best("TX", "heart attack")

[1] "CYPRESS FAIRBANKS MEDICAL CENTER"

best("TX", "heart failure")

[1] "FORT DUNCAN MEDICAL CENTER"

best("MD", "heart attack")

[1] "JOHNS HOPKINS HOSPITAL, THE"

best("MD", "pneumonia")

[1] "GREATER BALTIMORE MEDICAL CENTER"

best("BB", "heart attack")

[1] "invalid state"

best("NY", "hert attack")

[1] "invalid outcome"

Ranking hospitals by outcome in a state

Code

rankhospital <- function(state, outcome, num = "best") {
  
  if(!(state %in% hospitals$State)) {
    return("invalid state")
  }
  
  outcome <- tolower(outcome)
  outcome_col <- switch(outcome,
                        "heart attack" = "heart_attack",
                        "heart failure" = "heart_failure",
                        "pneumonia" = "pneumonia",
                        return("invalid outcome"))
  
  state_df <- hospitals %>%
    filter(State == state) %>%
    arrange(!!sym(outcome_col), hospital_name)
  
  if(num == "best") {
    state_df |>
      slice_min(!!sym(outcome_col), with_ties = FALSE, na_rm = T) |> 
      pull(hospital_name)
  }
  
  else if(num == "worst") {
    state_df |>
      slice_max(!!sym(outcome_col), with_ties = FALSE, na_rm = T) |> 
      pull(hospital_name)
  }
  
  else if(is.numeric(num)) {
    if(num > 0 && num <= nrow(state_df)) {
      return(state_df$hospital_name[num])
    }
    else {
      return(NA)
    }
  }
}

To match the requirements for handling ties and required conditions, we must sort the values by the outcome and hospital name columns followed by using na_rm and with_ties arguments in slice_min() and slice_max() which return, combined with pull(), the demanded rows.

Testing

rankhospital("TX", "heart failure", 4)

[1] "DETAR HOSPITAL NAVARRO"

rankhospital("MD", "heart attack", "worst")

[1] "HARFORD MEMORIAL HOSPITAL"

rankhospital("MN", "heart attack", 5000)

[1] NA

rankhospital("NY", "pneumonia", 10)

[1] "SISTERS OF CHARITY HOSPITAL"

Ranking hospitals in all states

Code

rankall <- function(outcome, num = "best"){
  
  outcome <- tolower(outcome)
  outcome_col <- switch(outcome,
                        "heart attack" = "heart_attack",
                        "heart failure" = "heart_failure",
                        "pneumonia" = "pneumonia",
                        return("invalid outcome"))
  
  state_df <- hospitals |>
    arrange(!!sym(outcome_col), hospital_name) |> 
    group_by(State)
  
  if(num == "best"){
    state_df |> 
      slice_min(!!sym(outcome_col), with_ties = FALSE, na_rm = T) |> 
      select(hospital = hospital_name, state = State)
  }
  
  else if(num == "worst"){
    state_df |> 
      slice_max(!!sym(outcome_col), with_ties = FALSE, na_rm = T) |> 
      select(hospital = hospital_name, state = State)
  }
  
  else if(is.numeric(num)) {
    if(num > 0) {
      state_df |> 
        slice_min(!!sym(outcome_col), 
                  n = num, 
                  with_ties = FALSE, 
                  na_rm = T) |> 
        slice(num) |> 
        select(hospital = hospital_name, state = State) |> 
        right_join(data.frame(state = unique(hospitals$State))) |> 
        arrange(state)
    }
    else{
      return("Invalid num")
    }
  }
}

Often solved with too many lines of code, the case of numeric values for the ranking is easier dealt with by using the n argument that controls the number of rows to display and slice() that retrieves them using the index which is the num value in this case.

However the function will ignore states having less hospitals than the demanded ranking (num argument), that’s why we use right_join() to add them back followed by arrange() to match the output sample in the assignment document.

Testing

tail(rankall("heart failure"), 10)

hospital	state
WELLMONT HAWKINS COUNTY MEMORIAL HOSPITAL	TN
FORT DUNCAN MEDICAL CENTER	TX
VA SALT LAKE CITY HEALTHCARE - GEORGE E. WAHLEN VA MEDICAL CENTER	UT
SENTARA POTOMAC HOSPITAL	VA
GOV JUAN F LUIS HOSPITAL & MEDICAL CTR	VI
SPRINGFIELD HOSPITAL	VT
HARBORVIEW MEDICAL CENTER	WA
AURORA ST LUKES MEDICAL CENTER	WI
FAIRMONT GENERAL HOSPITAL	WV
CHEYENNE VA MEDICAL CENTER	WY

tail(rankall("pneumonia", "worst"), 3)

hospital	state
MAYO CLINIC HEALTH SYSTEM - NORTHLAND, INC	WI
PLATEAU MEDICAL CENTER	WV
NORTH BIG HORN HOSPITAL DISTRICT	WY

head(rankall("heart attack", 20), 10)

hospital	state
NA	AK
D W MCMILLAN MEMORIAL HOSPITAL	AL
ARKANSAS METHODIST MEDICAL CENTER	AR
JOHN C LINCOLN DEER VALLEY HOSPITAL	AZ
SHERMAN OAKS HOSPITAL	CA
SKY RIDGE MEDICAL CENTER	CO
MIDSTATE MEDICAL CENTER	CT
NA	DC
NA	DE
SOUTH FLORIDA BAPTIST HOSPITAL	FL