Q_25.Rmd

---
title: "Q_2"
author: "Viet Minh Thong Le- Dinh Triem Phan"
date: "05/05/2020"
output: pdf_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```


```{r echo=FALSE, r,echo=F}
library(ggplot2)
library(gridExtra)
library(ggfortify)
library(leaps)
library(MASS)
library(ggplot2)
```

# Etude des données 


```{r}
Q2 = read.csv("Q_2.csv")
```


```{r}
Q2_1 = Q2[,c("A2a","A2b","A2c","A2d","A2e","A2f")]
summary(Q2_1)
gameper=apply(Q2_1[,c(1:6)],2,mean)
gameper[6]=mean(Q2_1$A2f,na.rm=TRUE)
print(gameper)
```



```{r}
Q2_5=Q2[,c("A1a","A1b","A1c","A1d","A2a","A2b","A2c","A2d","A2e","A2f","X5a","X5b","X5c","X5d")]
Q2_5[is.na(Q2_5)]=0
Q2_5[,'A1']=Q2_5[,'A1a']*0+Q2_5[,'A1b']*1+Q2_5[,'A1c']*2+Q2_5[,'A1d']*3
Q2_5$A2a=factor(Q2_5$A2a,levels=c(0,1),labels=c("No","Yes"))
Q2_5$A2b=factor(Q2_5$A2b,levels=c(0,1),labels=c("No","Yes"))
Q2_5$A2c=factor(Q2_5$A2c,levels=c(0,1),labels=c("No","Yes"))
Q2_5$A2d=factor(Q2_5$A2d,levels=c(0,1),labels=c("No","Yes"))
Q2_5$A2e=factor(Q2_5$A2e,levels=c(0,1),labels=c("No","Yes"))
Q2_5$A2f=factor(Q2_5$A2f,levels=c(0,1),labels=c("No","Yes"))
Q2_5$A1=factor(Q2_5$A1,levels=c(0,1,2,3),labels=c("once","less_5","5to10","more_10"))
Q2_5[,'A5']=Q2_5[,'X5a']*0+Q2_5[,'X5b']*1+Q2_5[,'X5c']*2+Q2_5[,'X5d']*3
Q2_5$A5=factor(Q2_5$A5,levels=c(0,1,2,3),labels=c("positive","negative","average","no_idea"))
Q2_5=Q2_5[,c(5:10,15,16)]
colnames(Q2_5)=c("Pendu","Audiowords","Check_your_motus","QCM","Check_your_taboo","Flashcards","frequency","satisfied")
summary(Q2_5)
```


```{r}
set.seed(100) # initialisation du générateur
# Extraction des échantillons
test.ratio=.2   # part de l'échantillon test
npop=nrow(Q2_5) # nombre de lignes dans les données
nvar=ncol(Q2_5) # nombre de colonnes
# taille de l'échantillon test
ntest=ceiling(npop*test.ratio) 
# indices de l'échantillon test
testi=sample(1:npop,ntest)
# indices de l'échantillon d'apprentissage
appri=setdiff(1:npop,testi)
```

```{r}

# construction de l'échantillon d'apprentissage
datappq=Q2_5[appri,]
# construction de l'échantillon test 
datestq=Q2_5[testi,] 
summary(datappq) # vérification
```




```{r echo=FALSE}
library(rpart)
library(partykit)
tree.q5=rpart(satisfied~.,data=datappq,cp=0.0001)
xmat5=xpred.rpart(tree.q5)
xerr5=(xmat5!=as.numeric(datappq[,"satisfied"]))^2
CVerr5=apply(xerr5,2,sum)
CVerr5
```




```{r echo=FALSE}
tree.dis5=rpart(satisfied~.,data=datappq,
                control=rpart.control(cp=as.numeric(attributes(which.min(CVerr5))$names)))
plot(as.party(tree.dis5), type="simple")
```

```{r}
# Calcul des prévisions
pred.treeq5=predict(tree.dis5,newdata=datestq,type="class") 
# Erreur quadratique moyenne de prévision en régression
sum((as.numeric(pred.treeq5)!=as.numeric(datestq[,"satisfied"])))/nrow(datestq)
```

```{r}
table(pred.treeq5,datestq[,"satisfied"])
```






```{r}
Q2_5t=Q2_5
Q2_5t[,"X1B"]=Q2[,"X1B"]

Q2_5t=Q2_5t[-which(is.na(Q2_5t$X1B)),]
colnames(Q2_5t)=c("Pendu","Audiowords","Check_your_motus","QCM","Check_your_taboo","Flashcards","frequency","satisfied","Time_CYS")
summary(Q2_5t)
```










```{r}
set.seed(320) # initialisation du générateur
# Extraction des échantillons
test.ratio=.2   # part de l'échantillon test
npop=nrow(Q2_5t) # nombre de lignes dans les données
nvar=ncol(Q2_5t) # nombre de colonnes
# taille de l'échantillon test
ntest=ceiling(npop*test.ratio) 
# indices de l'échantillon test
testi=sample(1:npop,ntest)
# indices de l'échantillon d'apprentissage
appri=setdiff(1:npop,testi)
```

```{r}

# construction de l'échantillon d'apprentissage
datappq=Q2_5t[appri,]
# construction de l'échantillon test 
datestq=Q2_5t[testi,] 
summary(datappq) # vérification
```





```{r}
mean(Q2_5t$Time_CYS)
hist(Q2_5t$Time_CYS,xlab="Time spent on CYS(hours)",ylab="Number of students",main="Histogram")
```



```{r}
tree.q5t=rpart(satisfied~.,data=datappq,cp=0.0001)
xmat5t=xpred.rpart(tree.q5t)
xerr5t=(xmat5t!=as.numeric(datappq[,"satisfied"]))^2
CVerr5t=apply(xerr5t,2,sum)
CVerr5t
```



```{r echo=FALSE}
tree.dis5t=rpart(satisfied~.,data=datappq,
                control=rpart.control(cp=as.numeric(attributes(which.min(CVerr5t))$names)))
plot(as.party(tree.dis5t), type="simple")
```