# STAT_540_Lec_09_R

# importing data into R

# get the working directory
getwd()

# set the working directory
setwd("/Users/karlgregory/Desktop/stat540_data")

# sometimes useful to list all the files in the working directory:
files <- list.files(getwd())
dat <- files[grep(".dat",files)] # list all files with ".dat" in the name

# on a windows machine change "/" to "\\"
safari <- read.table(file = "safari_comma.dat",
           sep = ",",
           header = T)

is.data.frame(safari)

# read in a data set with missing values and lines at the beginning to skip
safari <- read.table(file = "safari_comma_missing.dat",
                     sep = ",",
                     header = T,
                     skip = 3,
                     na.strings = c(".",""))
safari



# read in a data set with missing values and lines at the beginning to skip. Tab-delimited
safari <- read.table(file = "safari_tab_missing.dat",
                     sep = "\t",
                     header = T,
                     skip = 3,
                     na.strings = c(".",""))
safari


# read in fixed-width data

safari <- read.fwf(file = "safari_fwf_missing.dat",
                   header = T,
                   skip = 3,
                   na.strings = c(".",""),
                   sep = ",",
                   widths = c(11,3,5,3,14,9,9,4,18),
                   strip.white = T)

safari$date

day <- as.Date("1/13/1999",format = "%m/%d/%Y")
as.Date("1999-01-13") # default format

day

# strptime() function is for not only dates but times

strptime("1/13/1999", format= "%m/%d/%Y")

strptime("1/13/1999 10:30", format= "%m/%d/%Y %H:%M")

strptime("1/13/1999 10:30 pm", format= "%m/%d/%Y %I:%M %p")

strptime("1/13/1999 22:30", format= "%m/%d/%Y %H:%M")

my_time <- strptime("January 13, 1999 10:30 pm", format= "%B %d, %Y %I:%M %p")


# the strftime(), string format time, for putting a time into a string with a desired format
# make it like this: 1999.01.13 2230
strftime(my_time, format = "%Y.%m.%d %H%M")


# for the safari data: Change the date format to yyyy-mm-dd

dates <- strptime(safari$date,format="%m/%d/%Y")

safari$date <- strftime(dates,"%Y-%m-%d")

safari

# let's get the duration of each safari

start <- strptime(safari$start,format = "%I:%M %p")
end <- strptime(safari$end,format = "%I:%M %p")
duration <- difftime(end,start,units="min") # duration of each safari in minutes

# would like to show the durations in the format hh:mm
# we have to do it "manually"  :

hrs <- floor(duration / 60)
min <- as.numeric(duration) %% 60

# sprintf() a function for printing numbers in desired formats

sprintf(50, fmt = "%.03f")
sprintf(50, fmt = "%03.f")
sprintf(8, fmt = "%02.f")

paste(hrs,min,sep=":")

safari$duration <- paste(sprintf(hrs,fmt="%02.f"),sprintf(min,fmt="%02.f"),sep=":")

safari


# processing text


safari$weather

# grep() function. Search for a "pattern" within a string

sunny <- grep("sunny",safari$weather) # in which entries it found a match
safari$sunny <- FALSE
safari$sunny[sunny] <- TRUE


## How about changing a full name to a first initial and last name

# strsplit() split a string on a character
ch <- "Karl Bruce Gregory"


name_abb <- function(ch){

  full <- strsplit(ch," ")[[1]]
  n <- length(full)
  
  if(n == 1){
    
    abb <- full
    
  } else {
    
    abb <- paste(substr(full[1],1,1),". ",full[n], sep = "")
    
  }

  return(abb)
  
}

name_abb("Karl Bruce Gregory")
name_abb("Matthew Frederick Thaddeus Bailey")
name_abb("Bailey")

# use sapply to apply the function to the column in the data set
safari$name_abb <- sapply(safari$guide,name_abb)
safari

# sort_by for sorting data frames

sort_by(safari, ~ date)