STATISTICA SUI DATI DI GENERE - distribuzioni per sesso, età e cittadinanza per le 8 circoscrizione

eta_media.R

library('dplyr')
library('reshape2')
library('ggplot2')
library('Hmisc')

options(stringsAsFactors=FALSE) 

getFiles <- function(m) {

mydir<-"E:/github/ewpal"
temp<-tempfile(tmpdir=mydir, fileext=".zip")
download.file(m, temp)
unzip(temp, exdir=mydir)
unlink(temp) #delete the zip file

}

readFiles <- function(file) {

	t <- read.csv2(file, skip = 1, header = FALSE, 
		           stringsAsFactors = FALSE)

	b <- unlist(strsplit(file, " ",  fixed = TRUE))
	b <- sub(".csv", "", b)
	b <- data.frame(matrix(rep(b, each = nrow(t)), ncol = 4))

	x <- data.frame(t, b)
	return(x)
}

###############################################################################

files <- c("http://www.comune.palermo.it/js/server/uploads/opendata/_30072015131336.zip",
           "http://www.comune.palermo.it/js/server/uploads/opendata/_30072015131412.zip",
           "http://www.comune.palermo.it/js/server/uploads/opendata/_30072015131425.zip",
           "http://www.comune.palermo.it/js/server/uploads/opendata/_30072015131435.zip")

# Read files
# lapply(files, getFiles)

# Read dir
m <- dir()

x <- do.call("rbind", 
         lapply(as.list(m[grep("circ", m)]), readFiles))
names(x) <- c("eta", "1 circ", "2 circ", "3 circ", "4 circ", 
	                 "5 circ", "6 circ", "7 circ", "8 circ", 
	                 "n.d.", "anno", "sesso", "origine", "region")

z <- do.call("rbind", 
	     lapply(as.list(m[grep("quart", m)]), readFiles))

# löschen der Zeilen "n.d."

z1 <- melt(x, idvars=c("eta", "anno", "sesso", "origine", "region"), factorsAsStrings = FALSE)
z1$variable <- as.character(z1$variable)

str(z1)

sum(z1[z1$eta == "n.d.", "value"])

z2 <- tbl_df(z1) %>%
        filter(!eta %in% c("n.d.", "Totale")) %>%
          filter(!variable %in% c("n.d.")) %>%
         group_by(anno, sesso, variable) %>%
         summarise( pop       = sum(value),
         	        eta_media = wtd.mean(as.numeric(eta) + 0.5, value)) 

z2

ggplot(z2, aes(anno, eta_media, group = variable, color = variable)) + geom_line() + facet_wrap(~ sesso)
z2[z2$variable == "1 circ" & z2$sex == "maschi", ]

p1 <- ggplot(z2, aes(anno, eta_media)) + geom_boxplot(aes(fill = sesso))
p1 <- p1 + ggtitle('STATISTICA SUI DATI DI GENERE ANNO 2011 - 2014 - 8 circoscrizione // Età media')
p1 <- p1 + theme(plot.title = element_text(size=12, face="bold", vjust=1))
p1
ggsave(p1, file="eta_media_circo.pdf", units = "mm", width=297, height=210)

#
# FINI
#