Using sort and rank in R on multiple columns

问题

I’m trying to rank my hospital name by lowest rate for each state. When multiple hospitals have the same rate, the tie should be broken by using the hospital name and sorting it alphabetically. So far I’ve managed to rank it by rate within the state sorting it by hospital name, but I can’t figure out how to break the ties and rank it without skipping numbers

This is what I’ve got so far by using the following code:

outcome_data <- read.csv("outcome-of-care-measures.csv", na.strings="Not Available" ,stringsAsFactors=FALSE) #Read csv file
myData = outcome_data[,c(2, 7, 11)] #Retrieve only Hosp name, state and heart attack rate
arr1<-myData[complete.cases(myData[,3]),]  ##Remove NAs
arr2 <- arr1[order(arr1[2], arr1[3], arr1[1]),] #sort by state, then rate and then hospital name
arr3<-transform(arr2, rank = ave(rate, State, FUN = function(x) rank(x, ties.method = "min"))) #Rank by rate within each state

The output I currently get is:

Hospital.Name                           State  rate  rank
SOUTH PENINSULA HOSPITAL                AK     10.8  1
YUKON KUSKOKWIM DELTA REG HOSPITAL      AK     11.2  2
MAT-SU REGIONAL MEDICAL CENTER          AK     11.4  3
PEACEHEALTH KETCHIKAN MEDICAL CENTER    AK     11.4  3
ALASKA NATIVE MEDICAL CENTER            AK     11.6  5
BARTLETT REGIONAL HOSPITAL              AK     11.6  5
CENTRAL PENINSULA GENERAL HOSPITAL      AK     11.6  5
PROVIDENCE ALASKA MEDICAL CENTER        AK     12.4  8
ALASKA REGIONAL HOSPITAL                AK     13.4  9
FAIRBANKS MEMORIAL HOSPITAL             AK     15.6  10
GEORGE H. LANIER MEMORIAL HOSPITAL      AL     8.8   1
EVERGREEN MEDICAL CENTER                AL     9.1   2
BAPTIST MEDICAL CENTER EAST             AL     9.6   3
LAWRENCE MEDICAL CENTER                 AL     9.9   4
ANDALUSIA REGIONAL HOSPITAL             AL     10.1  5
JACKSON HOSPITAL & CLINIC INC           AL     10.2  6
BIRMINGHAM VA MEDICAL CENTER            AL     10.4  7
FLORALA MEMORIAL HOSPITAL               AL     10.4  7
GROVE HILL MEMORIAL HOSPITAL            AL     10.4  7
SPRINGHILL MEDICAL CENTER               AL     10.4  7
WEDOWEE HOSPITAL                        AL     10.4  7
PARKWAY MEDICAL CENTER                  AL     10.5  12
ST VINCENT'S BIRMINGHAM                 AL     10.6  13
WIREGRASS MEDICAL CENTER                AL     10.6  13
GADSDEN REGIONAL MEDICAL CENTER         AL     10.7  15
HALE COUNTY HOSPITAL                    AL     10.7  15
MOBILE INFIRMARY                        AL     10.7  15

But what I’m trying to get is

Hospital.Name                           State  rate  rank
SOUTH PENINSULA HOSPITAL                AK     10.8  1
YUKON KUSKOKWIM DELTA REG HOSPITAL      AK     11.2  2
MAT-SU REGIONAL MEDICAL CENTER          AK     11.4  3
PEACEHEALTH KETCHIKAN MEDICAL CENTER    AK     11.4  4
ALASKA NATIVE MEDICAL CENTER            AK     11.6  5
BARTLETT REGIONAL HOSPITAL              AK     11.6  6
CENTRAL PENINSULA GENERAL HOSPITAL      AK     11.6  7
PROVIDENCE ALASKA MEDICAL CENTER        AK     12.4  8
ALASKA REGIONAL HOSPITAL                AK     13.4  9
FAIRBANKS MEMORIAL HOSPITAL             AK     15.6  10
GEORGE H. LANIER MEMORIAL HOSPITAL      AL     8.8   1
EVERGREEN MEDICAL CENTER                AL     9.1   2
BAPTIST MEDICAL CENTER EAST             AL     9.6   3
LAWRENCE MEDICAL CENTER                 AL     9.9   4
ANDALUSIA REGIONAL HOSPITAL             AL     10.1  5
JACKSON HOSPITAL & CLINIC INC           AL     10.2  6
BIRMINGHAM VA MEDICAL CENTER            AL     10.4  7
FLORALA MEMORIAL HOSPITAL               AL     10.4  8
GROVE HILL MEMORIAL HOSPITAL            AL     10.4  9
SPRINGHILL MEDICAL CENTER               AL     10.4  10
WEDOWEE HOSPITAL                        AL     10.4  11
PARKWAY MEDICAL CENTER                  AL     10.5  12
ST VINCENT'S BIRMINGHAM                 AL     10.6  13
WIREGRASS MEDICAL CENTER                AL     10.6  14
GADSDEN REGIONAL MEDICAL CENTER         AL     10.7  15
HALE COUNTY HOSPITAL                    AL     10.7  16
MOBILE INFIRMARY                        AL     10.7  17

Any ideas?

回答1:

Using data.table this is relatively straight-forward:

library(data.table)

# Read only relevant columns from csv file using data.table::fread
outcome_data <- fread("outcome-of-care-measures.csv",
                      na.strings="Not Available" ,
                      select = c("Hospital.Name","State","rate"))

# Drop rows NA values using data.table::na.omit
outcome_data <- na.omit(outcome_data)

## Use data.table::setkey to sort/index by State, then rate, then hospital name
setkey(outcome_data,State,rate,Hospital.Name)

## Add a rank column by state, order within groups will be based key order above
## (the .N operator is the number of rows in each State group)
outcome_data[,rank := seq_len(.N),by = .(State)]

回答2:

We need a sequence number by group after the order step

library(dplyr)
arr2 %>%
     group_by(State) %>%
     mutate(rank = row_number())

Or if we are starting from 'arr1'

arr1 %>%
   arrange(State, rate,  Hospital.Name) %>%
   group_by(State) %>%
   mutate(rank = row_number())

Or using ave from base R

with(arr2, ave(seq_along(State), State, FUN = seq_along))
#[1]  1  2  3  4  5  6  7  8  9 10  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17

来源：https://stackoverflow.com/questions/48018499/using-sort-and-rank-in-r-on-multiple-columns

标签

sorting

rank