Why did she got an A while I got a D?

analysis code

#Dataset description
library(plyr); library(dplyr)
library(ggplot2)
setwd("~/Desktop/student")
d1_mat=read.csv("student-mat.csv", sep=';', header=T, stringsAsFactors = F)
d2_por=read.csv("student-por.csv", sep=';', header = T, stringsAsFactors = F)
data=merge(d1_mat,d2_por,by=c("school", "sex", "age", "address", "famsize", 
                              "Pstatus","Medu", "Fedu", "Mjob", "Fjob", "reason",
                              "nursery", "internet"))
summary(data)

#Which school has better student performance?
library(lattice)
library(plyr)
library(Rmisc)
p1<-ggplot(data=data, aes(x=school, y=G3.x))+
  geom_point(aes(color=school))+
  geom_boxplot(aes(color=school))+
  xlab('School')+
  ylab("Final Grade of Math")+
  geom_hline(yintercept=9,colour='grey20',size=0.5,linetype=2)+
  geom_hline(yintercept=11,colour='grey20',size=0.5,linetype=2)+
  geom_hline(yintercept=13,colour='grey20',size=0.5,linetype=2)+
  geom_hline(yintercept=15,colour='grey20',size=0.5,linetype=2)+
  annotate('text', x=2.5, y=c(8.5,10.5,12.5,14.5, 17), 
           label=c('F', 'D', 'C', 'B', 'A'), colour='red')+
  ggtitle("G3 in Math vs School")+
  theme_bw()+
  theme(legend.position="none")
p2<-ggplot(data=data, aes(x=school, y=G3.y))+
  geom_point(aes(color=school))+
  geom_boxplot(aes(color=school))+
  scale_x_discrete(name='School')+
  scale_y_continuous(name="Final Grade of Portuguese")+
  geom_hline(yintercept=9,colour='grey20',size=0.5,linetype=2)+
  geom_hline(yintercept=11,colour='grey20',size=0.5,linetype=2)+
  geom_hline(yintercept=13,colour='grey20',size=0.5,linetype=2)+
  geom_hline(yintercept=15,colour='grey20',size=0.5,linetype=2)+
  annotate('text', x=2.5, y=c(8.5,10.5,12.5,14.5, 17), 
           label=c('F', 'D', 'C', 'B', 'A'), colour='red')+
  ggtitle("G3 in Portuguese vs School")+
  theme_bw()+
  theme(legend.position="none")
multiplot(p1, p2, cols =1)

#Does the current student performance have a correlation with the past?
p3<-ggplot(data=data, aes(x=G1.x, y=G3.x, color=school))+
  geom_point(size=1, shape=1, position = "jitter")+
  scale_x_continuous(name="First Period Grade of Math")+
  scale_y_continuous(name="Final Grade of Math")+
  ggtitle("G3 vs G1 in Math")+
  scale_color_manual(values=c('red', 'blue'))+
  theme_bw()
p4<-ggplot(data=data, aes(x=G2.x, y=G3.x, color=school))+
  geom_point(size=1, shape=1, position = "jitter")+
  scale_x_continuous(name="Second Period Grade of Math")+
  scale_y_continuous(name="Final Grade of Math")+
  ggtitle("G3 vs G2 in Math")+
  scale_color_manual(values=c('red', 'blue'))+
  theme_bw()
multiplot(p3, p4, cols=2)

#Students who did not drop class:
data_G3xgreater0=data[data$G3.x>0,]
 p5<-ggplot(data=data_G3xgreater0, aes(x=G2.x, y=G3.x, color=school))+
     geom_point(size=1, shape=1, position = "jitter")+
     scale_x_continuous(name="Second Period Grade of Math")+
     scale_y_continuous(name="Final Grade of Math")+
     ggtitle("G3 vs G2 in Math")+
     scale_color_manual(values=c('red', 'green') )+
     geom_smooth(se=F, method='lm', aes(group=1), alpha=0.8)+
     theme_bw()
p6<-ggplot(data=data_G3xgreater0, aes(x=G1.x, y=G3.x, color=school))+
     geom_point(size=1, shape=1, position = "jitter")+
     scale_x_continuous(name="First Period Grade of Math")+
     scale_y_continuous(name="Final Grade of Math")+
     ggtitle("G3 vs G1 in Math")+
     scale_color_manual(values=c('red', 'green') )+
     geom_smooth(se=F, method='lm', aes(group=1), alpha=0.8)+
     theme_bw()
multiplot(p5, p6, cols=2)

#Students who did not drop the course:
data_G3x0<-data[data$G3.x==0,]
 p7<-ggplot(data=data_G3x0, aes(x=school, y=G1.x))+
     geom_point(aes(color=school))+
     geom_boxplot(aes(color=school))+
     scale_x_discrete(name='School')+
     scale_y_continuous(name='First Grade of Math')+
     geom_hline(yintercept=9,colour='grey20',size=0.5,linetype=2)+
  geom_hline(yintercept=11,colour='grey20',size=0.5,linetype=2)+
  geom_hline(yintercept=13,colour='grey20',size=0.5,linetype=2)+
  geom_hline(yintercept=15,colour='grey20',size=0.5,linetype=2)+
  annotate('text', x=2.5, y=c(8.5,10.5,12.5,14.5, 17), 
           label=c('F', 'D', 'C', 'B', 'A'), colour='red')+
     ggtitle('students who drop Grade 3')+
     theme_bw()+
   theme(legend.position="none")
p8<-ggplot(data=data_G3x0, aes(x=school, y=G2.x))+
     geom_point(aes(color=school))+
     geom_boxplot(aes(color=school))+
     scale_x_discrete(name='School')+
     scale_y_continuous(name='Second Grade of Math')+
   geom_hline(yintercept=9,colour='grey20',size=0.5,linetype=2)+
  geom_hline(yintercept=11,colour='grey20',size=0.5,linetype=2)+
  geom_hline(yintercept=13,colour='grey20',size=0.5,linetype=2)+
  geom_hline(yintercept=15,colour='grey20',size=0.5,linetype=2)+
  annotate('text', x=2.5, y=c(8.5,10.5,12.5,14.5, 17), 
           label=c('F', 'D', 'C', 'B', 'A'), colour='red')+
     ggtitle('students who drop Grade 3')+
     theme_bw()+
  theme(legend.position="none")
data_G2x0<-data[data$G2.x==0,]
multiplot(p7, p8,cols=2)

nrow(data_G2x0)
# 13 dropped G2
nrow(data_G3x0)
# 39 dropped G3
data_Gx00<-data[(data$G2.x==0 & data$G3.x==0),]
nrow(data_Gx00)
# 13 dropped both G2, G3

#Does student performance affect by past class failure?
p9<-ggplot(data=data, aes(x=failures.x, y=G3.x))+
  geom_point(aes(color=failures.x))+
  geom_boxplot(aes(group=failures.x, color=failures.x))+
  xlab('Past Class Failures')+
  scale_y_continuous(name='Final Grade of Math')+
  ggtitle('G3 vs Failures')+
  theme_bw()+
  facet_wrap(~school)+
  theme(legend.position="none")

p10<-ggplot(data=data, aes(x=failures.y, y=G3.y))+
  geom_point(aes(color=failures.y))+
  geom_boxplot(aes(group=failures.y, color=failures.y))+
  xlab('Past Class Failures')+
  scale_y_continuous(name='Final Grade of Portuguese')+
  theme_bw()+
  facet_wrap(~school)+
  theme(legend.position="none")
multiplot(p9, p10, cols=1)

#Does student performance change based on age?
p11<-ggplot(data=data, aes(x=age, y=G3.x))+
  geom_point(aes(color=age))+
  geom_boxplot(aes(group=age, color=age))+
  xlab('Age')+
  ylab('Final Grade of Math')+
  ggtitle('G3 vs Age')+
  theme_bw()+
  facet_wrap(~school)+
  theme(legend.position="none")
p12<-ggplot(data=data, aes(x=age, y=G3.y))+
  geom_point(aes(color=age))+
  geom_boxplot(aes(group=age, color=age))+
  xlab('Age')+
  ylab('Final Grade of Portuguese')+
  theme_bw()+
  facet_wrap(~school)+
  theme(legend.position="none")
multiplot(p11, p12, cols=1)

#Does the student who wants to take higher education do better at school?
p13<-ggplot(data=data, aes(x=higher.x, y=G3.x))+
  geom_point(aes(color=higher.x))+
  geom_boxplot(aes(color=higher.x))+
  scale_x_discrete(name='Wants to take higher education')+
  scale_y_continuous(name='Final Grade of Math')+
  ggtitle('G3 vs Higher')+
  theme_bw()+
  facet_grid(school~.)+
  theme(legend.position="none")
p14<-ggplot(data=data, aes(x=higher.y, y=G3.y))+
  geom_point(aes(color=higher.y))+
  geom_boxplot(aes(color=higher.y))+
  scale_x_discrete(name='Wants to take higher education')+
  scale_y_continuous(name='Final Grade of Portuguese')+
  theme_bw()+
  ggtitle('G3 vs Higher')+
  facet_grid(school~.)+
  theme(legend.position="none")
multiplot(p13, p14, cols=2)

#Is the fact true that the more time a student spend on studying, he has higher grade?
p15<-ggplot(data=data, aes(x=studytime.x, y=G3.x))+
  geom_point(aes(color=studytime.x))+
  geom_boxplot(aes(group=studytime.x, color=studytime.x))+
  xlab('Weekly Study Time')+
  scale_y_continuous(name='Final Grade of Math')+
  ggtitle('G3 vs Study Time')+
  theme_bw()+
  facet_wrap(~school)+
  theme(legend.position="none")
p16<-ggplot(data=data, aes(x=studytime.y, y=G3.y))+
  geom_point(aes(color=studytime.y))+
  geom_boxplot(aes(group=studytime.y, color=studytime.y))+
  xlab('Weekly Study Time')+
  scale_y_continuous(name='Final Grade of Portuguese')+
  theme_bw()+
  facet_wrap(~school)+
  theme(legend.position="none")
multiplot(p15,p16,cols=1)

#Does absence relate with student performance?
p17<-ggplot(data=data, aes(x=absences.x, y=G3.x))+
  geom_point(aes(color=absences.x))+
  geom_boxplot(aes(group=absences.x, color=absences.x))+
  xlab('Absences')+
  ylab('Final Grade of Math')+
  ggtitle('G3 vs Absences')+
  facet_grid(school~.)+
  theme_bw()+
  scale_color_gradient(low='red', high='yellow', name='absences')+
  coord_cartesian(xlim=c(0, 30))+
  theme(legend.position="none")
p18<-ggplot(data=data, aes(x=absences.y, y=G3.y))+
  geom_point(aes(color=absences.y))+
  geom_boxplot(aes(group=absences.y, color=absences.y))+
  xlab('Absences')+
  ylab('Final Grade of Portuguese')+
  facet_grid(school~.)+
  theme_bw()+
  scale_color_gradient(low='red', high='yellow', name='absences')+
  coord_cartesian(xlim=c(0, 30))+
  theme(legend.position="none")
multiplot(p17, p18, cols=1)

#To get a better understanding of the plot, I grouped number of school absences to 4 categories: 0-9, 10-19, 20-29, 30+.
data$absences.xgroup=numeric(nrow(data))
for (i in 1:nrow(data)){
  if (data$absences.x[i]>=0&data$absences.x[i]<=9){
    data$absences.xgroup[i]="0-9"
  }
  else if (data$absences.x[i]>=10&data$absences.x[i]<=19){
    data$absences.xgroup[i]="10-19"
  }
  else if (data$absences.x[i]>=20&data$absences.x[i]<=29){
    data$absences.xgroup[i]="20-29"
  }
  else if (data$absences.x[i]>=15&data$absences.x[i]<=19){
    data$absences.xgroup[i]="15-19"
  }
  else
    data$absences.xgroup[i]="30+"
}
data$absences.ygroup=numeric(nrow(data))
for (i in 1:nrow(data)){
  if (data$absences.y[i]>=0&data$absences.y[i]<=9){
    data$absences.ygroup[i]="0-9"
  }
  else if (data$absences.y[i]>=10&data$absences.y[i]<=19){
    data$absences.ygroup[i]="10-19"
  }
  else if (data$absences.y[i]>=20&data$absences.y[i]<=29){
    data$absences.ygroup[i]="20-29"
  }
  else if (data$absences.y[i]>=15&data$absences.y[i]<=19){
    data$absences.ygroup[i]="15-19"
  }
  else
    data$absences.ygroup[i]="30+"
}
p19<-ggplot(data=data, aes(x=absences.xgroup, y=G3.x))+
  geom_point(aes(color=absences.xgroup))+
  geom_boxplot(aes(group=absences.xgroup, color=absences.xgroup))+
  xlab('Absences Group')+
  ylab('Final Grade of Math')+
  ggtitle('G3 vs Absences Group')+
  facet_grid(school~.)+
  theme_bw()+
  theme(legend.position="none")
p20<-ggplot(data=data, aes(x=absences.ygroup, y=G3.y))+
  geom_point(aes(color=absences.ygroup))+
  geom_boxplot(aes(group=absences.ygroup, color=absences.ygroup))+
  xlab('Absences Group')+
  ylab('Final Grade of Math')+
  ggtitle('G3 vs Absences Group')+
  facet_grid(school~.)+
  theme_bw()+
  theme(legend.position="none")
multiplot(p19, p20, cols=2)

#Does the parents' education and job influence student performance?
p21<-ggplot(data=data, aes(x=Mjob, y=G3.x))+
  geom_point(aes(color=Mjob))+
  xlab('Mother Job')+
  ylab('Final Grade of Math')+
  ggtitle('G3 vs Mother Job')+
  geom_boxplot(aes(group=Mjob, color=Mjob))+
  facet_grid(school~.)+
  theme_bw()+
  theme(legend.position="none")
p22<-ggplot(data=data, aes(x=Mjob, y=G3.y))+
  geom_point(aes(color=Mjob))+
  xlab('Mother Job')+
  ylab('Final Grade of Portuguese')+
  geom_boxplot(aes(group=Mjob, color=Mjob))+
  facet_grid(school~.)+
  theme_bw()+
  theme(legend.position="none")
  p23<-ggplot(data=data, aes(x=Medu, y=G3.x))+
  geom_point(aes(color=Medu))+
  xlab('Mother Education')+
  ylab('Final Grade of Math')+
  ggtitle('G3 vs Mother Education')+
  geom_boxplot(aes(group=Medu, color=Medu))+
  facet_grid(school~.)+
  theme_bw()+
  theme(legend.position="none")
p24<-ggplot(data=data, aes(x=Medu, y=G3.y))+
    geom_point(aes(color=Medu))+
    xlab('Mother Education')+
    ylab('Final Grade of Portuguese')+
    geom_boxplot(aes(group=Medu, color=Medu))+
    facet_grid(school~.)+
    theme_bw()+
  theme(legend.position="none")
data$Meducha=as.character(data$Medu)
p25<-ggplot(data=data, aes(x=Mjob))+
  geom_bar(aes(fill=Meducha), position='fill')+
  xlab('Mother Job')+
  ggtitle('Mjob vs Medu')+
  theme_bw()
multiplot(p21, p22,p25, p23, p24, cols=2)