European Parliament Election 2019 - Provincial Results
Giovanni Valentini
Saturday, July 20, 2019
Our purpose is to analyse the results of the European Parliament election in Italy for the year 2019. The analysis will be performed at provincial level.
library(dplyr)
library(ggplot2)
setwd("~/Coursera/Data_Science")
fileloc <- "./datasets/Italia/europee2019_scrutini_area_italia.csv"
itdata <- read.csv(fileloc, header = TRUE, sep = ";", strip.white = TRUE, stringsAsFactors = FALSE)
names(itdata)[c(3,4,5,6)] <- c("Provincia", "Comune", "Lista", "Voti")
str(itdata)## 'data.frame': 124515 obs. of 6 variables:
## $ CIRCOSCRIZIONE: chr "I : ITALIA NORD-OCCIDENTALE" "I : ITALIA NORD-OCCIDENTALE" "I : ITALIA NORD-OCCIDENTALE" "I : ITALIA NORD-OCCIDENTALE" ...
## $ REGIONE : chr "LIGURIA" "LIGURIA" "LIGURIA" "LIGURIA" ...
## $ Provincia : chr "GENOVA" "GENOVA" "GENOVA" "GENOVA" ...
## $ Comune : chr "ARENZANO" "ARENZANO" "ARENZANO" "ARENZANO" ...
## $ Lista : chr "PARTITO DEMOCRATICO" "LEGA SALVINI PREMIER" "MOVIMENTO 5 STELLE" "FORZA ITALIA" ...
## $ Voti : int 1904 1809 969 415 350 236 150 115 60 33 ...1 - Analysis at provincial level
provdata <- itdata %>%
group_by(CIRCOSCRIZIONE, REGIONE, Provincia, Lista) %>%
summarise(Voti = sum(Voti))
prov_total <- provdata %>%
group_by(Provincia) %>%
summarise(TotVoti = sum(Voti))
print(arrange(prov_total, desc(TotVoti)))## Source: local data frame [107 x 2]
##
## Provincia TotVoti
## (chr) (int)
## 1 ROMA 1723583
## 2 MILANO 1452603
## 3 TORINO 1117522
## 4 NAPOLI 1027989
## 5 BRESCIA 643829
## 6 BERGAMO 579830
## 7 BARI 527280
## 8 FIRENZE 521666
## 9 BOLOGNA 517039
## 10 PADOVA 487405
## .. ... ...In the following table we display the vote percentages given to the first party, that is LEGA, in each of the 18 provinces of the 5 Italian regions selected (Abruzzo, Lazio, Marche, Molise, Umbria):
## compute the vote percentages for each Party (i.e. Lista)
provdata <- inner_join(x = provdata, y = prov_total, by = "Provincia")
provdata <- mutate(provdata, pctVoti = round(x = 100 * Voti / TotVoti, digits = 2))
## display the vote percentages for Lega (The League)
nameLista <- "LEGA SALVINI PREMIER"
nameRegioni <- c("ABRUZZO", "LAZIO", "MARCHE", "MOLISE", "UMBRIA")
mydata <- provdata[provdata$Lista == nameLista & provdata$REGIONE %in% nameRegioni, c("REGIONE", "Provincia", "Voti", "pctVoti")]
print(arrange(mydata, desc(pctVoti)))## Source: local data frame [18 x 4]
##
## REGIONE Provincia Voti pctVoti
## (chr) (chr) (int) (dbl)
## 1 MARCHE FERMO 35194 43.37
## 2 LAZIO VITERBO 64810 41.42
## 3 LAZIO RIETI 31573 41.04
## 4 MARCHE MACERATA 64063 40.99
## 5 LAZIO FROSINONE 96671 40.35
## 6 LAZIO LATINA 91230 38.90
## 7 UMBRIA TERNI 42031 38.56
## 8 ABRUZZO TERAMO 51515 38.13
## 9 UMBRIA PERUGIA 129427 38.06
## 10 MARCHE ASCOLI PICENO 41155 38.00
## 11 MARCHE PESARO E URBINO 72898 37.82
## 12 ABRUZZO L'AQUILA 43471 35.67
## 13 ABRUZZO CHIETI 58727 34.42
## 14 MARCHE ANCONA 77751 34.13
## 15 ABRUZZO PESCARA 51657 33.54
## 16 LAZIO ROMA 509605 29.57
## 17 MOLISE CAMPOBASSO 27349 24.57
## 18 MOLISE ISERNIA 9195 23.382 - Plot of the vote percentages for the 3 main parties
mycol <- c("blue", "yellow", "tomato")
parties <- c("LEGA SALVINI PREMIER", "MOVIMENTO 5 STELLE", "PARTITO DEMOCRATICO")
mydata <- provdata[provdata$Lista %in% parties & provdata$REGIONE %in% nameRegioni, ]
# Boxplot
gg <- ggplot(data = mydata, aes(x = Lista, y = pctVoti, fill = Lista))
gg + geom_boxplot() + scale_fill_manual(values = mycol)
# Barplot
gg <- ggplot(data = mydata, aes(x = Provincia, y = pctVoti, fill = Lista))
gg + geom_bar(stat = "identity", position = "dodge") + scale_fill_manual(values = mycol) + coord_flip()
3 - Scatterplot of the vote percentages for 2 parties
Now we want to compare the vote percentages of 2 parties, LEGA and M5S, across the 18 provinces:
parties <- c("LEGA SALVINI PREMIER", "FRATELLI D'ITALIA", "FORZA ITALIA",
"MOVIMENTO 5 STELLE", "PARTITO DEMOCRATICO")
mydata <- provdata[provdata$Lista %in% parties & provdata$REGIONE %in% nameRegioni, ]
# first we need to spread rows into columns
library(tidyr)
wdata <- spread(data = mydata[, -c(5:6)], key = Lista, value = pctVoti)
print(wdata[, -(1:2)])## Source: local data frame [18 x 6]
##
## Provincia FORZA ITALIA FRATELLI D'ITALIA LEGA SALVINI PREMIER
## (chr) (dbl) (dbl) (dbl)
## 1 FROSINONE 8.13 8.90 40.35
## 2 LATINA 11.98 10.58 38.90
## 3 RIETI 6.68 8.17 41.04
## 4 ROMA 5.92 8.88 29.57
## 5 VITERBO 6.35 8.55 41.42
## 6 ANCONA 4.88 5.40 34.13
## 7 ASCOLI PICENO 6.55 5.92 38.00
## 8 FERMO 5.42 5.64 43.37
## 9 MACERATA 5.87 7.91 40.99
## 10 PESARO E URBINO 5.50 4.66 37.82
## 11 PERUGIA 6.52 6.70 38.06
## 12 TERNI 6.10 6.22 38.56
## 13 CHIETI 9.36 5.63 34.42
## 14 L'AQUILA 10.42 10.54 35.67
## 15 PESCARA 9.90 6.72 33.54
## 16 TERAMO 7.92 5.87 38.13
## 17 CAMPOBASSO 13.00 5.47 24.57
## 18 ISERNIA 21.84 8.75 23.38
## Variables not shown: MOVIMENTO 5 STELLE (dbl), PARTITO DEMOCRATICO (dbl)rownames(wdata) <- wdata$Provincia
names(wdata) <- gsub(pattern = " ", replacement = ".", x = names(wdata))
gs <- ggplot(data = wdata, aes(x = LEGA.SALVINI.PREMIER, y = MOVIMENTO.5.STELLE))
gs + geom_point(aes(text = paste("Regione:", REGIONE)), size = 2, color = "forestgreen")
4 - Principal Components Analysis (PCA)
pcx <- princomp(x = wdata[, -c(1:3)], cor = FALSE)
summary(pcx)## Importance of components:
## Comp.1 Comp.2 Comp.3 Comp.4 Comp.5
## Standard deviation 7.0403320 4.1627923 3.1840604 1.19978052 0.33169730
## Proportion of Variance 0.6307518 0.2205168 0.1290134 0.01831791 0.00140009
## Cumulative Proportion 0.6307518 0.8512686 0.9802820 0.99859991 1.00000000plot(pcx) # screeplot: Variance vs PC
loadings(pcx)##
## Loadings:
## Comp.1 Comp.2 Comp.3 Comp.4 Comp.5
## FORZA.ITALIA 0.489 -0.304 0.440 -0.479 0.495
## FRATELLI.D'ITALIA -0.181 0.389 0.836 0.341
## LEGA.SALVINI.PREMIER -0.697 -0.516 -0.226 -0.166 0.412
## MOVIMENTO.5.STELLE 0.437 -0.762 0.177 0.442
## PARTITO.DEMOCRATICO -0.291 0.779 0.151 -0.111 0.523
##
## Comp.1 Comp.2 Comp.3 Comp.4 Comp.5
## SS loadings 1.0 1.0 1.0 1.0 1.0
## Proportion Var 0.2 0.2 0.2 0.2 0.2
## Cumulative Var 0.2 0.4 0.6 0.8 1.0biplot(pcx, col = c(4, 2), cex = c(0.6, 0.7), xlabs = rownames(wdata))
5 - Map of the results
library(maps)
mymap <- map_data(map = "italy")
mymap$Provincia <- toupper(mymap$region)
area <- inner_join(x = provdata, y = mymap, by = "Provincia")
## choose a party (Lista)
votiLega <- area[area$Lista == "LEGA SALVINI PREMIER" & area$REGIONE %in% nameRegioni, ]
summary(votiLega$pctVoti)## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 23.38 34.13 38.06 36.09 38.90 41.42## define 5 intervals of votes percentages
votiLega$pctVoti_LEGA <- cut_interval(x = votiLega$pctVoti, n = 5, dig.lab = 3)
gg <- ggplot() +
geom_polygon(data = votiLega, aes(x = long, y = lat, group = group, fill = pctVoti_LEGA), color = "blue")
gg + scale_fill_brewer(palette = "PuBu") + coord_fixed(ratio = 1.3) + theme_bw()