## ----CAP394_PCDDataScience_Read01----------------------------------------
file = "Data/PCDs/PCD_20060101_20061231.dat"
# Let's read the file, using spaces as separators, with the first line as the header, 
# read only 10 rows, missing strings will be represented as NAs, 
# strings will not be factorized.
data <- read.table(file, sep = "", 
                    header = TRUE, nrows = 10,
                    na.strings ="NA", stringsAsFactors = FALSE)


## ----CAP394_PCDDataScience_Read02----------------------------------------
str(data)


## ----CAP394_PCDDataScience_Read03----------------------------------------
# Let's read one line of the file, expecting it to contain strings,
# without messages to the console.
scannedNames <- scan(text = readLines(file, 1), 
                     what="",quiet = TRUE)
# Let's remove the first element of the resulting vector.
scannedNames <- scannedNames[-1]
scannedNames


## ----CAP394_PCDDataScience_Read04----------------------------------------
# Let's read the file, using spaces as separators, ignoring the header, 
# missing strings will be represented as NAs, strings will not be factorized.
data <- read.table(file, sep = "", header = FALSE, 
                   na.strings ="NA", stringsAsFactors = FALSE, col.names = scannedNames)


## ----CAP394_PCDDataScience_Read05----------------------------------------
colnames(data) <- c("station","longitude","latitude","altitude","relhumidity",
                    "maxtemp","mintemp","rainfall","date")


## ----CAP394_PCDDataScience_Read05s---------------------------------------
str(data)


## ----CAP394_PCDDataScience_Read05s2--------------------------------------
# Get the header of a subset of the data frame containing only 
# station, maxtemp, mintemp, rainfall, date.
head(subset(data, select = c(station,maxtemp,mintemp,rainfall,date)))


## ----CAP394_PCDDataScience_Clean01---------------------------------------
# Select all columns but longitude,latitude,altitude
data <- subset(data,select=-c(longitude,latitude,altitude))
str(data)


## ----CAP394_PCDDataScience_Clean02---------------------------------------
# Get the date column as a character, convert it to date using the format YYYYMMDD.
data$date <- as.Date(as.character(data$date), "%Y%m%d")


## ----CAP394_PCDDataScience_Clean03---------------------------------------
# Replace all 999s in column relhumidity
data$relhumidity[data$relhumidity == 999] <- NA
# Replace all 9999.99s in column maxtemp
data$maxtemp[data$maxtemp == 9999.99] <- NA
# Replace all 9999.99s in column mintemp
data$mintemp[data$mintemp == 9999.99] <- NA
# Replace all 9999.99s in column rainfall
data$rainfall[data$rainfall == 9999.99] <- NA
str(data)


## ----CAP394_PCDDataScience_Question01------------------------------------
# Get the minimum and maximum value of the respective fields, disregarding NAs.
min(data$mintemp,na.rm = TRUE)
max(data$maxtemp,na.rm = TRUE)


## ----CAP394_PCDDataScience_Question02------------------------------------
# Get the rows where the mintemp is the minimum.
# This does not work: data[data$mintemp==-5,] because it will also consider 
# NA values 
# https://stackoverflow.com/questions/1686569/filter-data-frame-rows-by-a-logical-condition
subset(data, mintemp==min(data$mintemp,na.rm = TRUE))
# Do the same for maxtemp.
subset(data, maxtemp==max(data$maxtemp,na.rm = TRUE))


## ----CAP394_PCDDataScience_Question03------------------------------------
# Create new deltatemp column by an operation on the mintemp and maxtemp.
data$deltatemp = data$maxtemp-data$mintemp


## ----CAP394_PCDDataScience_Question04------------------------------------
subset(data, deltatemp==max(data$deltatemp,na.rm = TRUE))


## ----CAP394_PCDDataScience_Viz01,fig.width=12,fig.height=6---------------
# Select only the variables relevant to the plot: date (x axis), mintemp (y axis)
# and station (grouping)
subset <- subset(data,select=c(date,station,mintemp))
# ggplot2 requires the data frame to be reshaped, using data and station as ids.
# https://stackoverflow.com/questions/13324004/plotting-multiple-time-series-in-ggplot
library(reshape2)
melted <- melt(subset,id=c("date","station"))
# Now we can plot the multiple time series.
library(ggplot2)
ggplot(melted,aes(x=date,y=value,colour=station,group=station)) + geom_line()


## ----CAP394_PCDDataScience_Viz02,fig.width=12,fig.height=6---------------
subset2 <- subset[subset$station %in% c("32492","32565","32548","32595","32465","31998"), ]
melted <- melt(subset2,id=c("date","station"))
# Plot the three time series.
ggplot(melted,aes(x=date,y=value,colour=as.factor(station),group=station)) + geom_line()


## ----CAP394_PCDDataScience_Viz03,fig.width=12,fig.height=6---------------
subset3 <- subset[subset$station %in% c("31998"), ]
melted <- melt(subset3,id=c("date","station"))
# Plot the three time series.
ggplot(melted,aes(x=date,y=value,colour=as.factor(station),group=station)) + geom_line()