Graduation Rates

setwd("/Users/tt/Documents")

grad <-read.csv("grad-rate2.csv", colClasses="character")
graddat <- data.frame(grad)
head(graddat)

##                      School     City..State X4.year.graduation.rate
## 1          Juilliard School    New York, NY                     94%
## 2           Bowdoin College   Brunswick, ME                     91%
## 3 College of the Holy Cross   Worcester, MA                     91%
## 4           Amherst College     Amherst, MA                     90%
## 5            Babson College Babson Park, MA                     90%
## 6     Georgetown University  Washington, DC                     90%

names(graddat)

## [1] "School"                  "City..State"            
## [3] "X4.year.graduation.rate"

graddat$rate <- as.numeric(gsub("%","",graddat$X4.year.graduation.rate))
str(graddat)

## 'data.frame':    1249 obs. of  4 variables:
##  $ School                 : chr  "Juilliard School" "Bowdoin College" "College of the Holy Cross" "Amherst College" ...
##  $ City..State            : chr  "New York, NY" "Brunswick, ME" "Worcester, MA" "Amherst, MA" ...
##  $ X4.year.graduation.rate: chr  "94%" "91%" "91%" "90%" ...
##  $ rate                   : num  94 91 91 90 90 90 90 90 90 90 ...

#graddat<- graddat[,c(1,4,6,7)]
graddat$n_chars = nchar(graddat$City..State)
str(graddat$n_chars)

##  int [1:1249] 12 13 13 11 15 14 11 15 13 14 ...

graddat$city = as.character(substr(graddat$City..State, 1, graddat$n_chars-4))
graddat$state = as.character(substr(graddat$City..State, graddat$n_chars-1, graddat$n_chars))
hist(graddat[,4],breaks= 100, xlab = "Graduation Rates", main="Top Graduation Rates by State", col = 'skyblue',border = 'white')

## [1] "School"                  "City..State"            
## [3] "X4.year.graduation.rate"
graddat2<- graddat[,c(1,4,7,6)]

best<-function(state1){
stdat <-data.frame(graddat2)
if (!state1 %in% unique(stdat$state)){stop("invalid state")}
stdf <- stdat[stdat$state == state1,]
stdf <- na.omit(stdf)
max_stdf<- stdf[which.max(stdf[,2]),1]
print(max_stdf)
}
worst<-function(state1){
stdat <-data.frame(graddat2)
if (!state1 %in% unique(stdat$state)){stop("invalid state")}
stdf <- stdat[stdat$state == state1,]
stdf <- na.omit(stdf)
min_stdf<- stdf[which.min(stdf[,2]),1]
print(min_stdf)
}

rankcollege<- function(state2, num="best"){
bsdat <-data.frame(graddat2)
if (!state2 %in% unique(bsdat$state)){stop("invalid state")}
bsdf <- bsdat[bsdat$state == state2,]
bsdf <- na.omit(bsdf)
ncollege<-nrow(bsdf)
switch(num, "best" = {num = ncollege},  "worst" = {num = 1} )
if (num > ncollege) {result = NA}
order_bsdf <- order(bsdf[,2], bsdf[ ,1])
rank_bsdf <- bsdf[order_bsdf,  ][num ,1]
rank_bsdf 
}
rank_allstates <- function(num="best") {   
  alldat <- data.frame(graddat2)
  allstates <- unique(alldat[, 3])
  print(head(allstates))
  alldat <- na.omit(alldat)
 in_state_rank<- function(state3){
 in_statedf <- alldat[alldat[,3] == state3,]
 mcollege<-nrow(in_statedf)
 switch(num, "best" = {num = mcollege},  "worst" = {num = 1} )
 if (num > mcollege) {result = NA}
 order_in_statedf <- order(in_statedf[,2], in_statedf[ ,1])
 result <- in_statedf[order_in_statedf,  ][num ,1]
 in_state<- c(result, state3)
 head(in_state,3)
  }
output = do.call(rbind, lapply(allstates, in_state_rank))
output = output[order(output[ ,2]), ]
row.names(output) = output[ ,2]
colnames(output) = c("college", "state")
data.frame(output)
 }

best("TX")

## [1] "Rice University"

best("NY")

## [1] "Juilliard School"

worst("TX")

## [1] "University of Houston--Downtown"

worst("NY")

## [1] "CUNY--Medgar Evers College"

rankcollege("NY",4)

## [1] "CUNY--City College"

rankcollege("NY","best")

## [1] "Juilliard School"

rankcollege("NY","worst")

## [1] "CUNY--Medgar Evers College"

rankcollege("TX",10)

## [1] "Texas A&M University--Kingsville"

rankcollege("TX","best")

## [1] "Rice University"

rankcollege("TX","worst")

## [1] "University of Houston--Downtown"

head(rank_allstates("best"), 10)

## [1] "NY" "ME" "MA" "DC" "CA" "MD"

##                            college state
## AK University of Alaska--Fairbanks    AK
## AL              Samford University    AL
## AR                 Hendrix College    AR
## AZ           University of Arizona    AZ
## CA      Soka University of America    CA
## CO                Colorado College    CO
## CT                 Yale University    CT
## DC           Georgetown University    DC
## DE          University of Delaware    DE
## FL             University of Miami    FL

tail(rank_allstates("worst"), 3)

## [1] "NY" "ME" "MA" "DC" "CA" "MD"

##                              college state
## WI University of Wisconsin--Parkside    WI
## WV    West Virginia State University    WV
## WY             University of Wyoming    WY

head(rank_allstates(2), 10)

## [1] "NY" "ME" "MA" "DC" "CA" "MD"

##                                           college state
## AK                University of Alaska--Anchorage    AK
## AL Alabama Agricultural and Mechanical University    AL
## AR             University of Arkansas--Pine Bluff    AR
## AZ Embry-Riddle Aeronautical University--Prescott    AZ
## CA         California State University--Fullerton    CA
## CO        Metropolitan State University of Denver    CO
## CT           Western Connecticut State University    CT
## DC         University of the District of Columbia    DC
## DE                          Wilmington University    DE
## FL                     Bethune-Cookman University    FL