Session2-ex1

Reading From Multiple Sources
Data Mining Stops
Frequency of Long Stops
Leaflet Package
ggplot2 group feauture
Grid Feauture
facet_wrap Feauture

Reading From Multiple Sources

#theData<-NULL
#for (i in 1:100) {
#  filename <- paste0(i,".txt")
#  info = file.info(filename)

###############check if it is an empty file before running codes

#  if(info$size != 0){
#    myData <- read.csv(filename, header = FALSE)
#    theData <- rbind(theData, myData)
#  }
# }

#colnames(theData) <- c("ID", "Date", "Lon", "Lat")
#theData <- as.data.frame(theData)
#theData$Date <- as.POSIXct(theData$Date)

Data Mining Stops

#trips <-NULL
#thisRow =theData[1,]
######create a new file 'trips' to aggregate useless lines 
#for (i in 2:length(theData$Lat2)) {
#  if(thisRow$Lat2 != theData$Lat2[i] | thisRow$Lon2 != theData$Lon2[i] | thisRow$ID != theData$ID[i]){
#    waitingTime <- as.integer(difftime(theData$Date[i-1], thisRow$Date, units = "mins"))
#    trips <- rbind(trips, cbind(theData[i-1,], waitingTime))
#    thisRow =theData[i,]
#  }
#}

#write.csv(trips, "trips100.csv", row.names = FALSE)

Frequency of Long Stops

trips <- read.csv("./data/trips100.csv")
head(trips[trips$waitingTime>0,])

##    ID                Date      Lon      Lat   Lon2  Lat2 waitingTime
## 16  1 2008-02-03 10:30:29 116.6929 39.85167 116.69 39.85         769
## 30  1 2008-02-04 10:05:10 116.6916 39.85171 116.69 39.85         889
## 35  1 2008-02-04 11:05:09 116.5094 39.90679 116.51 39.91          10
## 39  1 2008-02-05 09:38:34 116.6916 39.85156 116.69 39.85         749
## 49  1 2008-02-05 11:18:34 116.4609 39.91620 116.46 39.92           5
## 74  1 2008-02-05 18:51:20 116.5034 39.90693 116.50 39.91          10

dim(trips)

## [1] 60154     7

longStops <- trips[trips$waitingTime>=4*60,]
longStopsFreq <- aggregate( waitingTime~ Lat2+ Lon2, data=longStops, FUN=length)
colnames(longStopsFreq)[3] <- "num"
longStopsFreq <- longStopsFreq[longStopsFreq$num>5,]
longStopsFreq <- longStopsFreq[order(longStopsFreq$num, decreasing = TRUE),]

head(longStopsFreq)

##      Lat2   Lon2 num
## 98  40.20 116.72  24
## 95  39.72 116.71  14
## 80  40.11 116.66  12
## 71  40.13 116.64  10
## 97  40.13 116.72  10
## 102 40.15 116.73   9

Leaflet Package

For more INFO see https://rstudio.github.io/leaflet/

longStopsFreq1 <- data.frame(lng1=(longStopsFreq$Lon2-0.005),lat1=(longStopsFreq$Lat2-0.005),
                             lng2=(longStopsFreq$Lon2+0.005),lat2=(longStopsFreq$Lat2+0.005)) 


dim(longStopsFreq1)

## [1] 21  4

library(leaflet) 
m <- leaflet() %>%
  addTiles() %>%
addRectangles(longStopsFreq1$lng1,longStopsFreq1$lat1,longStopsFreq1$lng2, longStopsFreq1$lat2,
              popup  = paste0("Number of records: ",longStopsFreq$num),
              label  = paste0("Number of records: ",longStopsFreq$num))
m

#popup: hover and show
#label: click and show

ggplot2 group feauture

library(lubridate)
library(ggplot2)
days=c("", "Sat","Sun", "Mon", "Tue","Wed", "Thu", "Fri")

movingFreq <- aggregate(ID~day(Date)+hour(Date),data=trips[trips$waitingTime<60,], FUN=length)
colnames(movingFreq) <- c("day","hour", "num")

j <- ggplot(movingFreq, aes(x=hour))+
  geom_line(aes(y =num, group=day, colour=days[day]))+
  geom_point(aes(y =num, group=day, colour=days[day]))+
  labs(title ="Comparision of Frequency Working T Drive Taxis", 
       x = "Hour", y = "Frequency", col="Day")+
  theme(panel.background = element_rect(fill = 'white'))+
  scale_color_brewer(palette="Paired")
print(j)

Grid Feauture

library(gridExtra)
library(grid)
weekendsFreq<-movingFreq[movingFreq$day<4,]
k1 <- ggplot(weekendsFreq, aes(x=hour))+
  geom_line(aes(y =num, group=day, colour=days[day]))+
  geom_point(aes(y =num, group=day, colour=days[day]))+
  labs(x = "Hour", y = "Frequency", col="Day")+
  theme(panel.background = element_rect(fill = 'white'))

weekdaysFreq<-movingFreq[movingFreq$day>=4,]
k2 <- ggplot(weekdaysFreq, aes(x=hour))+
  geom_line(aes(y =num, group=day, colour=days[day]))+
  geom_point(aes(y =num, group=day, colour=days[day]))+
  labs(x = "Hour", y = "Frequency", col="Day")+
  theme(panel.background = element_rect(fill = 'white'))

grid.arrange(
  k1,
  k2,
  nrow = 1,
  top = "Comparision of Frequency Working T Drive Taxis",
  bottom = textGrob(
    "Fig 3",
    gp = gpar(fontface = 3, fontsize = 9),
    hjust = 1,
    x = 1
  )
)

facet_wrap Feauture

nineIDs<-trips[trips$ID %in% c(1,2,4,5,7,11,13,20,21), ]

p1<-ggplot(nineIDs, aes(x=Lat))+
  geom_path(aes(y =Lon))

p4 <-
  p1 + facet_wrap( ~ ID, nrow = 3) + theme(legend.position = "none") +
  ggtitle("Visited Places by 9 Taxi Drivers")
p4