Labos en R

In [ ]:

    

# help()
# help(function)
# help(package='package-name)


    

In [ ]:

    

# install
# install.packages('package-name')

# already installed with conda
#install.packages("foreign")

# new installs
#install.packages("Rcmdr", dependencies = TRUE, repos="http://cran.rstudio.com/") # in conda?
#install.packages("nortest", repos="http://cran.rstudio.com/")
#install.packages("sas7bdat", repos="http://cran.rstudio.com/")
#install.packages("Hmisc", repos="http://cran.rstudio.com/")
#install.packages("pastecs", repos="http://cran.rstudio.com/")
# import
# library('package-name')

library(foreign)
library(nortest)
library(sas7bdat)
library(Hmisc)
library(pastecs)


    

In [ ]:

    

ls()
# rm(list=ls())
# setwd()
getwd()


    

In [ ]:

    

# import excel : via txt tab separated
#fichierTexte <- read.table("data/labo2/SR_Data.txt", header = TRUE)

# import DBF (DBase)
fichierDBF <- read.dbf("data/labo2/SR_Data.dbf")

# import SPSS
#fichierSPSS <- read.spss("data/labo2/Data_SPSS.sav", to.data.frame=TRUE)

# import SAS
#fichierSAS <- read.sas7bdat("data/labo2/tableau1.sas7bdat", debug=FALSE)

head(fichierDBF)


    

In [ ]:

    

# show variable names
names(fichierDBF)
# indexes start at 1

# delete variable
fichierDBF$Shape_Leng <- NULL

# rename variable
names(fichierDBF)[1] <- "POPTOT"

# create variable
fichierDBF$km <- fichierDBF$Shape_Area / 1000000
fichierDBF$HabKm2 <- fichierDBF$POPTOT / fichierDBF$km

head(fichierDBF)


    

In [ ]:

    

# new table from a subset
names(fichierDBF)
ZScores <-fichierDBF[,c(12:15)]
names(ZScores)


    

In [ ]:

    

#install.packages("moments", repos="http://cran.rstudio.com/")
library(moments)


    

In [ ]:

    

skewness(fichierDBF)


    

In [ ]:

    

kurtosis(fichierDBF)


    

In [ ]:

    

#lillie.test(Tableau1$HabKm2)
# sapply(fichierDBF[18:20],lillie.test)
sapply(fichierDBF[18],lillie.test)


    

In [ ]:

    

#ks.test(x, y) # two sample

#m <- mean(fichierDBF[18])
#s <- sd(fichierDBF[18])

#ks.test(fichierDBF[18], "pnorm", m, s)


    

In [ ]:

    

sapply(fichierDBF[18],shapiro.test)  # sapply(fichierDBF[18:20],shapiro.test)


    

In [ ]:

    

fichierDBF$SqrtDens <- sqrt(fichierDBF$HabKm2)
fichierDBF$SqrtImg <- sqrt(fichierDBF$IMMREC_PCT)


    

In [ ]:

    

# log(0) = error
fichierDBF$LogDens <- log(fichierDBF$HabKm2)
fichierDBF$LogImg <- log(fichierDBF$IMMREC_PCT+1)

summary(fichierDBF)


    

In [ ]:

    

ZScores$INDICE_PAU <- scale(fichierDBF[1], center = TRUE, scale = TRUE)
ZScores$Dist_Min <- scale(fichierDBF[2], center = TRUE, scale = TRUE)
ZScores$N_1000 <- scale(fichierDBF[3], center = TRUE, scale = TRUE)
ZScores$Dist_Moy_3 <- scale(fichierDBF[4], center = TRUE, scale = TRUE)

#help(sapply)
sapply(ZScores,mean)
sapply(ZScores,sd)


    

In [ ]:

    

summary(fichierDBF)


    

In [ ]:

    

sapply(fichierDBF, mean)
sapply(fichierDBF, sd)
sapply(fichierDBF, min)
sapply(fichierDBF, max)
sapply(fichierDBF, median)
sapply(fichierDBF, range)
sapply(fichierDBF, quantile)


    

In [ ]:

    

# Hmisc.describe
describe(fichierDBF)


    

In [ ]:

    

# pastecs.stat.desc
stat.desc(fichierDBF, basic=TRUE, norm=TRUE)


    

In [ ]:

    

hist(fichierDBF$HabKm2, main="Histogramme", xlab="Habitants au km2", ylab="Effectif", breaks=10, col='lightblue')


    

In [ ]:

    

hist(fichierDBF$SqrtDens, main="Histogramme", xlab="Habitants au km2 (racine)", ylab="Effectif", breaks=10, col='gold')


    

In [ ]:

    

hist(fichierDBF$LogDens, main="Histogramme", xlab="Habitants au km2 log)", ylab="Effectif", breaks=10, col='coral')


    

In [ ]:

    

x <- fichierDBF$HabKm2
h<-hist(x, breaks=10, col="lightblue", xlab="Habitants au km2", ylab="Effectif", 
main="Histogramme avec courbe normale")
xfit<-seq(min(x),max(x),length=40)
yfit<-dnorm(xfit,mean=mean(x),sd=sd(x))
yfit <- yfit*diff(h$mids[1:2])*length(x)
lines(xfit, yfit, col="blue", lwd=2)


    

In [ ]:

    

x <- fichierDBF$SqrtDens
h<-hist(x, breaks=10, col="red", xlab="Habitants au km2 (racine)", ylab = "Effectif",
main="Histogramme avec courbe normale")
xfit<-seq(min(x),max(x),length=40)
yfit<-dnorm(xfit,mean=mean(x),sd=sd(x))
yfit <- yfit*diff(h$mids[1:2])*length(x)
lines(xfit, yfit, col="blue", lwd=2)


    

In [ ]:

    

# install
#install.packages('doBy', repos="http://cran.rstudio.com/")
#install.packages('gmodels', repos="http://cran.rstudio.com/")
#install.packages('scatterplot3d', repos="http://cran.rstudio.com/")

# import
library(foreign)
library(nortest)
library(sas7bdat)
library(Hmisc)
library(pastecs)
library(ggplot2)
library(doBy)
library(gmodels)
library(scatterplot3d)

# data
Tableau1 <- read.sas7bdat("data/labo3/tableau1.sas7bdat", debug=FALSE)
names(Tableau1)

TableauKhi2 <- read.sas7bdat("data/labo3/khi2.sas7bdat", debug=FALSE)
names(TableauKhi2)


    

In [ ]:

    

hist(Tableau1$IMMREC_PCT, breaks=10, xlab="Immigrants récents (%)", ylab = "Effectif", main="Histogramme")


    

In [ ]:

    

hist(Tableau1$IMMREC_PCT, breaks=20, xlab="Immigrants récents (%)", ylab = "Effectif", main="Histogramme")


    

In [ ]:

    

hist(Tableau1$IMMREC_PCT, density=20, breaks=20, xlab="Immigrants récents (%)", ylab = "Effectif", main="Histogramme")


    

In [ ]:

    

hist(Tableau1$IMMREC_PCT, breaks=20, col="red", xlab="Immigrants récents (%)", ylab = "Effectif", main="Histogramme") 
hist(Tableau1$IMMREC_PCT, breaks=20, col="lightyellow", xlab="Immigrants récents (%)", ylab = "Effectif", main="Histogramme") 
hist(Tableau1$IMMREC_PCT, breaks=20, col="lightsalmon", xlab="Immigrants récents (%)", ylab = "Effectif", main="Histogramme") 
hist(Tableau1$IMMREC_PCT, breaks=20, col="lightgreen", xlab="Immigrants récents (%)", ylab = "Effectif", main="Histogramme")


    

In [ ]:

    




    

In [ ]:

    

plot(
    hist(Tableau1$IMMREC_PCT, breaks=20),
    ylim=c(0, 80), col="lightgreen", xlab="Immigrants récents (%)", ylab = "Effectif", main="Histogramme"
)


    

In [ ]:

    

hist(Tableau1$IMMREC_PCT, col="lightgray", breaks=20, xlab="Immigrants récents (%)", ylab = "Proportion", main="Histogramme", prob=TRUE)


    

In [ ]:

    

m <- mean(Tableau1$IMMREC_PCT)
std <- sd(Tableau1$IMMREC_PCT)
hist(Tableau1$IMMREC_PCT, col="lightyellow", breaks=20, prob=TRUE, xlab="Immigrants récents (%)", ylab = "Proportion", main="Histogramme avec la courbe normale")
curve(dnorm(x, mean=m, sd=std), col="darkblue", lwd=2, add=TRUE)


    

In [ ]:

    

x <- Tableau1$IMMREC_PCT
h<-hist(x, breaks=20, col="lightyellow", xlab="Immigrants récents (%)", ylab = "Effectif", main="Histogramme avec la courbe normale") 
xfit<-seq(min(x),max(x),length=40) 
yfit<-dnorm(xfit,mean=mean(x),sd=sd(x)) 
yfit <- yfit*diff(h$mids[1:2])*length(x) 
lines(xfit, yfit, col="darkblue", lwd=2)


    

In [ ]:

    

plot(Tableau1$IMMREC_PCT, Tableau1$FAIBREVPCT, xlab="Immigrants récents (%)", ylab = "Faible revenu (%)", main="Nuage de points")


    

In [ ]:

    

plot(Tableau1$IMMREC_PCT, Tableau1$FAIBREVPCT, xlab="Immigrants récents (%)", ylab = "Faible revenu (%)", main="Nuage de points avec droite de régression")
abline(lsfit(Tableau1$IMMREC_PCT, Tableau1$FAIBREVPCT))


    

In [ ]:

    

pairs(~MONOPCT+MENAGE1PCT+TX_CHOM+FAIBREVPCT,data=Tableau1, 
      main="Matrice de nuages de points")


    

In [ ]:

    

scatterplot3d(Tableau1$MONOPCT, Tableau1$TX_CHOM, Tableau1$FAIBREVPCT, main="Nuage de points 3D")
scatterplot3d(Tableau1$MONOPCT, Tableau1$TX_CHOM, Tableau1$FAIBREVPCT, main="Nuage de points 3D", xlab="Familles monoparentales (%)", ylab="Taux de chômage", zlab="Faible revenu (%)");


    

In [ ]:

    

rcorr(cbind(Tableau1$MONOPCT,Tableau1$MENAGE1PCT,Tableau1$TX_CHOM,Tableau1$FAIBREVPCT,Tableau1$Dist_Min,Tableau1$N_1000), type="pearson")


    

In [ ]:

    

rcorr(cbind(Tableau1$MONOPCT,Tableau1$MENAGE1PCT,Tableau1$TX_CHOM,Tableau1$FAIBREVPCT,Tableau1$Dist_Min,Tableau1$N_1000), type="spearman")


    

In [ ]:

    

reg <- lm(TX_CHOM ~ FAIBREVPCT, data = Tableau1)
summary(reg)

names(Tableau1)


    

In [ ]:

    

names(TableauKhi2)


    

In [ ]:

    

# sex
table(TableauKhi2$SEX)
TableauKhi2$SEX <- factor(TableauKhi2$SEX, levels = c(1,2), labels = c("Homme", "Femme"))
table(TableauKhi2$SEX)

# transport mode
table(TableauKhi2$Mode)
TableauKhi2$Mode <- factor(TableauKhi2$Mode, levels = c(0:4), labels = c("Auto (conducteur)", "Auto (passager)", "Transport en commun", "Tranport actif", "Autres"))
table(TableauKhi2$Mode)

# distance
table(TableauKhi2$DIST)
TableauKhi2$DIST <- factor(TableauKhi2$DIST, levels = c(1:7), labels = c("Moins de 5 km", "5 à 9,9 km","10 à 14,9 km", "15 à 19,9 km", "20 à 24,9 km", "25 à 29,9 km", "30 km et plus"))
table(TableauKhi2$DIST)


    

In [ ]:

    

CrossTable(TableauKhi2$SEX, TableauKhi2$Mode, chisq=TRUE, expected=TRUE, resid=TRUE, format="SPSS")
CrossTable(TableauKhi2$SEX, TableauKhi2$DIST, chisq=TRUE, expected=TRUE, resid=TRUE, format="SPSS")
CrossTable(TableauKhi2$Mode, TableauKhi2$DIST, chisq=TRUE, expected=TRUE, resid=TRUE, format="SPSS")


    

In [ ]:

    

# import
library(foreign)
library(nortest)
library(sas7bdat)
library(doBy)

# data
MTL <- read.sas7bdat("data/labo4/mtl_ttest.sas7bdat", debug=FALSE)
TOR <- read.sas7bdat("data/labo4/tor_ttest.sas7bdat", debug=FALSE)
VAN <- read.sas7bdat("data/labo4/van_ttest.sas7bdat", debug=FALSE)
TROISRMR <- read.sas7bdat("data/labo4/troisrmr_anova.sas7bdat", debug=FALSE)
names(MTL)
names(TOR)
names(VAN)
names(TROISRMR)


    

In [ ]:

    

# modalités (labels)
table(MTL$SEX)
table(TOR$SEX)
table(VAN$SEX)
MTL$SEX <- factor(MTL$SEX, levels = c(1,2), labels = c("Homme", "Femme"))
TOR$SEX <- factor(TOR$SEX, levels = c(1,2), labels = c("Homme", "Femme"))
VAN$SEX <- factor(VAN$SEX, levels = c(1,2), labels = c("Homme", "Femme"))
table(MTL$SEX)
table(TOR$SEX)
table(VAN$SEX)

TROISRMR$CMA <- factor(TROISRMR$CMA, levels = c(462,535,933), labels = c("Montréal", "Toronto", "Vancouver"))
table(TROISRMR$CMA)


    

In [ ]:

    

var.test(TOTINC ~ SEX, alternative='two.sided', conf.level=.95, data=MTL)


    

In [ ]:

    

t.test(TOTINC~SEX, alternative='two.sided', conf.level=.95, var.equal=FALSE, data=MTL)


    

In [ ]:

    

t.test(TOTINC~SEX, alternative='two.sided', conf.level=.95, var.equal=TRUE, data=MTL)
boxplot(TOTINC~SEX, data = MTL, col = "coral", main="Boites à moustache (RMR de Montréal)", xlab="Sexe", ylab="Revenu total")
boxplot(LogTotInc~SEX, data = MTL, col = "coral", main="Boites à moustache (RMR de Montréal)", xlab="Sexe", ylab="Revenu total (log)")


    

In [ ]:

    

# doBy
summaryBy(GROSRT ~ CMA, TROISRMR, FUN=c(mean), na.rm=TRUE)


    

In [ ]:

    

boxplot(GROSRT ~ CMA, data = TROISRMR, col = "lightyellow", main="Boites à moustache", xlab="Région métropolitaine", ylab="Loyer ($)")  #Analyse de variance : test F


    

In [ ]:

    

anova.aov <- aov(GROSRT ~ CMA, data = TROISRMR)
summary(anova.aov)


    

In [ ]:

    

f_theorique <- qf(0.99, 2, 8379)
f_theorique
# qt() pour table Student t pour coefficient de ... 
# (voir autres cours)


    

In [ ]:

    

anova.r2 <- lm(GROSRT ~ CMA, data = TROISRMR)
summary(anova.r2)


    

In [ ]:

    

TukeyHSD(anova.aov)


    

In [ ]:

    

# install
#install.packages("MASS", repos="http://cran.rstudio.com/")      ## Tests de normalité supp.
#install.packages("car", repos="http://cran.rstudio.com/")      ## Companion to Applied Regression


    

In [4]:

    

# import
library(foreign)
library(MASS)
library(sas7bdat)
library(pastecs)
library(car)

# data
MTL <- read.sas7bdat("data/labo5/pauvretemtl.sas7bdat", debug=FALSE)
names(MTL)


    

    
Out[4]:





	
'srnom'
	
'FAIBREVPCT'
	
'SqrtChom'
	
'MONOPCT'
	
'menage1per'
	
'SqrtImmig'
	
'pasecol1524'
	
'tpspartiel'

In [5]:

    

# stats univariees
summary(MTL)


    

    
Out[5]:





     srnom       FAIBREVPCT        SqrtChom        MONOPCT     
 0001.00:  1   Min.   : 1.232   Min.   :0.000   Min.   : 0.00  
 0002.00:  1   1st Qu.:19.761   1st Qu.:2.568   1st Qu.:16.05  
 0003.00:  1   Median :28.699   Median :2.925   Median :21.23  
 0004.00:  1   Mean   :29.982   Mean   :2.995   Mean   :21.38  
 0005.00:  1   3rd Qu.:39.803   3rd Qu.:3.416   3rd Qu.:26.18  
 0006.00:  1   Max.   :82.642   Max.   :6.887   Max.   :51.28  
 (Other):500                                                   
   menage1per       SqrtImmig      pasecol1524      tpspartiel   
 Min.   : 3.943   Min.   :0.000   Min.   : 0.00   Min.   :30.65  
 1st Qu.:28.587   1st Qu.:1.454   1st Qu.:24.51   1st Qu.:41.26  
 Median :38.598   Median :1.961   Median :32.66   Median :45.47  
 Mean   :37.674   Mean   :2.074   Mean   :32.66   Mean   :45.61  
 3rd Qu.:46.752   3rd Qu.:2.543   3rd Qu.:40.93   3rd Qu.:49.65  
 Max.   :72.632   Max.   :5.078   Max.   :68.75   Max.   :69.79  
                                                                 

In [11]:

    

# regression lineaire multiple
ols <- lm(
    FAIBREVPCT ~ 
    SqrtChom + 
    MONOPCT + 
    menage1per + 
    SqrtImmig + 
    pasecol1524 + 
    tpspartiel, 
    data=MTL)
summary(ols)


    

    
Out[11]:





Call:
lm(formula = FAIBREVPCT ~ SqrtChom + MONOPCT + menage1per + SqrtImmig + 
    pasecol1524 + tpspartiel, data = MTL)

Residuals:
    Min      1Q  Median      3Q     Max 
-20.405  -4.290  -0.586   3.498  34.343 

Coefficients:
             Estimate Std. Error t value Pr(>|t|)    
(Intercept) -36.10961    2.20183 -16.400  < 2e-16 ***
SqrtChom      7.97771    0.57049  13.984  < 2e-16 ***
MONOPCT       0.60825    0.04703  12.933  < 2e-16 ***
menage1per    0.20132    0.02517   7.997 8.94e-15 ***
SqrtImmig     2.58049    0.37394   6.901 1.57e-11 ***
pasecol1524   0.10986    0.02818   3.899  0.00011 ***
tpspartiel    0.27767    0.05336   5.204 2.86e-07 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 6.326 on 499 degrees of freedom
Multiple R-squared:  0.8059,	Adjusted R-squared:  0.8035 
F-statistic: 345.2 on 6 and 499 DF,  p-value: < 2.2e-16

In [10]:

    

## coefficients standardisés
CoefStand <- lm(
    scale(FAIBREVPCT) ~ 
    scale(SqrtChom) + 
    scale(MONOPCT) + 
    scale(menage1per) + 
    scale(SqrtImmig) + 
    scale(pasecol1524) + 
    scale(tpspartiel), 
    data = MTL
)
summary(CoefStand)


    

    
Out[10]:





Call:
lm(formula = scale(FAIBREVPCT) ~ scale(SqrtChom) + scale(MONOPCT) + 
    scale(menage1per) + scale(SqrtImmig) + scale(pasecol1524) + 
    scale(tpspartiel), data = MTL)

Residuals:
     Min       1Q   Median       3Q      Max 
-1.42971 -0.30057 -0.04107  0.24509  2.40635 

Coefficients:
                     Estimate Std. Error t value Pr(>|t|)    
(Intercept)        -1.646e-16  1.971e-02   0.000  1.00000    
scale(SqrtChom)     3.899e-01  2.788e-02  13.984  < 2e-16 ***
scale(MONOPCT)      3.343e-01  2.585e-02  12.933  < 2e-16 ***
scale(menage1per)   1.825e-01  2.282e-02   7.997 8.94e-15 ***
scale(SqrtImmig)    1.714e-01  2.484e-02   6.901 1.57e-11 ***
scale(pasecol1524)  9.386e-02  2.407e-02   3.899  0.00011 ***
scale(tpspartiel)   1.268e-01  2.436e-02   5.204 2.86e-07 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.4433 on 499 degrees of freedom
Multiple R-squared:  0.8059,	Adjusted R-squared:  0.8035 
F-statistic: 345.2 on 6 and 499 DF,  p-value: < 2.2e-16

In [12]:

    

## multicolinéarité?
### Valeurs de VIF
vif(ols)
vif(ols) > 5 # problème de multicolinéarité (VIF > 5)?


    

    
Out[12]:





	
SqrtChom
		
1.99767242206301
	
MONOPCT
		
1.7169166500618
	
menage1per
		
1.33888028140516
	
SqrtImmig
		
1.58542766716237
	
pasecol1524
		
1.48920558001708
	
tpspartiel
		
1.52538045714979








    
Out[12]:





	
SqrtChom
		
FALSE
	
MONOPCT
		
FALSE
	
menage1per
		
FALSE
	
SqrtImmig
		
FALSE
	
pasecol1524
		
FALSE
	
tpspartiel
		
FALSE

In [13]:

    

## Graphiques et distance de cook
opar <- par(mfrow = c(2, 2), oma = c(0, 0, 1.1, 0))
plot(ols, las = 1)
par(opar)


    

In [14]:

    

## Histogramme sur les résidus et vérification de la normalité des résidus
m <- mean(residuals(ols))
std <- sd(residuals(ols))
hist(residuals(ols), col="lightyellow", breaks=20, prob=TRUE, xlab="Résidus OLS", ylab = "Proportion", main="Histogramme avec la courbe normale")
curve(dnorm(x, mean=m, sd=std), col="darkblue", lwd=2, add=TRUE)
stat.desc(residuals(ols), basic=TRUE, norm=TRUE)


    

    
Out[14]:





	
nbr.val
		
506
	
nbr.null
		
0
	
nbr.na
		
0
	
min
		
-20.4047979910667
	
max
		
34.3433376307782
	
range
		
54.748135621845
	
sum
		
-7.40241201668823e-14
	
median
		
-0.586204938060121
	
mean
		
-1.46830543304696e-16
	
SE.mean
		
0.27955974003622
	
CI.mean.0.95
		
0.549243372142321
	
var
		
39.5457460140541
	
std.dev
		
6.28854084935878
	
coef.var
		
-42828560787309032
	
skewness
		
0.676033517801706
	
skew.2SE
		
3.11329159768941
	
kurtosis
		
2.13343255710525
	
kurt.2SE
		
4.92204145621073
	
normtest.W
		
0.972936774972511
	
normtest.p
		
4.69730411019801e-08








    

In [16]:

    

# Repérer les valeurs aberrantes : Distance de Cook > 4 / n ou 8 / n
nobs <- NROW(na.omit(residuals(ols)))            # Nombre d'observations dans le jeu de données
cook <- cooks.distance(ols)                      # Distance de Cook
ypredit <- fitted.values(ols)                    # Y prédits par le modèle
res <- residuals(ols)                            # résidus (Y - Y prédits par le modèle)
res_std <- rstandard(ols)                        # résidus standardisés

a <- cbind(MTL, cook, ypredit, res, res_std)
a <- a[order(-cook), ]
a[cook > 4/nobs, ]    # Observations dont la distance de Cook > 4 / n
a[cook > 8/nobs, ]    # Observations dont la distance de Cook > 8 / n


    

    
Out[16]:





srnom
FAIBREVPCT
SqrtChom
MONOPCT
menage1per
SqrtImmig
pasecol1524
tpspartiel
cook
ypredit
res
res_std

	
21
0021.00
62.77603
4.053217
39.70588
45.04792
1.369306
52.63158
49.62963
0.006716149
52.54256
10.23347
1.631841
	
71
0073.00
55.39906
3.67194
25
55.08475
1.928971
58.82353
50
0.006426216
44.80397
10.59509
1.687955
	
114
0114.00
24.43325
3.635362
16.32653
37.07865
3.58418
16.94915
53.42466
0.005897266
36.23329
-11.80005
-1.876161
	
135
0133.00
45.43689
2.983635
14.94253
56.81818
2.944575
27.86885
51.52439
0.005863305
33.18773
12.24916
1.946705
	
373
0395.01
58.20896
3.708991
38.46154
42.0354
1.552301
44.0678
47.25275
0.005672756
47.30444
10.90451
1.735029
	
473
0601.02
33.96226
4.051199
24.84848
36
3.681114
14
45.74899
0.004762004
42.31183
-8.349568
-1.33217
	
484
0605.01
28.87324
2.43975
15.85366
34.4
1.453505
26.66667
35.31746
0.004694459
16.40964
12.4636
1.978397
	
494
0610.06
40.14337
3.132303
27.81457
21.86047
1.466471
37.3494
38.22394
0.004607817
28.69945
11.44392
1.81777
	
105
0106.00
17.71429
1.591115
28.125
42.37288
2.320008
43.47826
39.31624
0.004509052
23.90192
-6.187635
-0.9936036
	
20
0019.00
54.01302
3.445598
33.98058
49.38272
1.137147
53.62319
47.47475
0.004492352
43.99694
10.01608
1.593039
	
165
0164.00
27.43902
3.558617
8.695652
57.5
3.157497
30.43478
51.75439
0.004426104
35.0076
-7.568571
-1.208992
	
367
0385.00
16.31944
2.522625
13.04348
61.34021
1.950956
15.38462
47.2
0.004403388
24.1289
-7.80946
-1.246638
	
43
0044.00
55.58252
3.837797
24.28571
50.84746
2.246793
31.81818
51.73913
0.004392712
43.1758
12.40672
1.968917
	
143
0141.00
17.95666
2.761036
21.79487
46.28571
2.360668
34.48276
51.33929
0.004330848
32.62783
-14.67118
-2.325591
	
136
0134.00
28.14371
2.649065
28.57143
52.94118
2.182179
45.45455
51.30435
0.004211983
37.93137
-9.787662
-1.55654
	
66
0067.00
44.82759
2.790059
29.26829
40.8805
2.618615
23.68421
47.82609
0.004111846
34.82063
10.00695
1.590789
	
23
0023.00
50
3.281651
29.62963
45.21739
1.894055
65
45.83333
0.003924724
41.95125
8.048745
1.282856
	
18
0017.00
48.25
3.611576
35.78947
43.71859
0
41.93548
52.23881
0.003781063
42.38543
5.864572
0.9407829
	
368
0390.00
28.04878
3.735437
25.37313
44.44444
1.084652
28
39.55224
0.003407594
34.92921
-6.880426
-1.098303
	
166
0165.00
44.2623
3.441236
14.77273
45.29915
2.862992
41.55844
57.62712
0.003208054
37.404
6.858298
1.094223
	
132
0130.00
53.6
4.427445
8.888889
67.52336
4.697417
7.555556
69.79167
0.00209248
50.54303
3.05697
0.4973239
	
185
0184.00
27.03151
3.251772
27.54491
40.78947
1.815184
43.75
39.32927
0.001673101
35.20933
-8.177823
-1.297175
	
204
0199.00
28.35249
3.517752
21.53846
46.69118
2.354768
36.2069
42.36111
0.001505015
36.27156
-7.919066
-1.255955
	
292
0286.00
43.01288
3.711843
21.9697
29.17706
3.732667
27.04918
43.54839
0.001429056
37.43548
5.5774
0.8872145
	
400
0430.00
17.68642
2.256851
18.18182
45.69288
1.796906
45.66929
41.72185
0.001215297
23.39228
-5.705859
-0.9065917
	
334
0325.02
26.84015
3.508565
22.28261
33.10345
2.585823
28.57143
41.92593
0.0009907295
33.5519
-6.711754
-1.064183
	
278
0273.00
22.39521
2.947197
19.82379
42.15686
2.444119
34.14634
42.47788
0.0008003792
29.80063
-7.405418
-1.17297
	
342
0329.00
22.26833
2.512107
20.72539
37.9822
1.739579
47.82609
38.70192
0.0001845368
24.67387
-2.405544
-0.3819289
	
422
0512.02
10.38186
2.132007
13.93443
26.5861
1.760377
22.68908
40.93686
0.0001670965
13.12932
-2.747458
-0.4356323
	
202
0197.00
28.34994
2.40906
24.03433
48.54167
1.994467
42.85714
45.61028
8.764326e-05
30.02045
-1.670502
-0.2652086
	
433
0520.02
20.56396
2.769892
18.99038
15.15789
2.431183
29.24528
41.03832
2.907333e-05
21.47213
-0.9081725
-0.1442568









    
Out[16]:





srnom
FAIBREVPCT
SqrtChom
MONOPCT
menage1per
SqrtImmig
pasecol1524
tpspartiel
cook
ypredit
res
res_std

	
21
0021.00
62.77603
4.053217
39.70588
45.04792
1.369306
52.63158
49.62963
0.006716149
52.54256
10.23347
1.631841
	
71
0073.00
55.39906
3.67194
25
55.08475
1.928971
58.82353
50
0.006426216
44.80397
10.59509
1.687955
	
114
0114.00
24.43325
3.635362
16.32653
37.07865
3.58418
16.94915
53.42466
0.005897266
36.23329
-11.80005
-1.876161
	
373
0395.01
58.20896
3.708991
38.46154
42.0354
1.552301
44.0678
47.25275
0.005672756
47.30444
10.90451
1.735029
	
484
0605.01
28.87324
2.43975
15.85366
34.4
1.453505
26.66667
35.31746
0.004694459
16.40964
12.4636
1.978397
	
105
0106.00
17.71429
1.591115
28.125
42.37288
2.320008
43.47826
39.31624
0.004509052
23.90192
-6.187635
-0.9936036
	
20
0019.00
54.01302
3.445598
33.98058
49.38272
1.137147
53.62319
47.47475
0.004492352
43.99694
10.01608
1.593039
	
165
0164.00
27.43902
3.558617
8.695652
57.5
3.157497
30.43478
51.75439
0.004426104
35.0076
-7.568571
-1.208992
	
367
0385.00
16.31944
2.522625
13.04348
61.34021
1.950956
15.38462
47.2
0.004403388
24.1289
-7.80946
-1.246638
	
143
0141.00
17.95666
2.761036
21.79487
46.28571
2.360668
34.48276
51.33929
0.004330848
32.62783
-14.67118
-2.325591
	
66
0067.00
44.82759
2.790059
29.26829
40.8805
2.618615
23.68421
47.82609
0.004111846
34.82063
10.00695
1.590789
	
23
0023.00
50
3.281651
29.62963
45.21739
1.894055
65
45.83333
0.003924724
41.95125
8.048745
1.282856
	
18
0017.00
48.25
3.611576
35.78947
43.71859
0
41.93548
52.23881
0.003781063
42.38543
5.864572
0.9407829
	
368
0390.00
28.04878
3.735437
25.37313
44.44444
1.084652
28
39.55224
0.003407594
34.92921
-6.880426
-1.098303
	
132
0130.00
53.6
4.427445
8.888889
67.52336
4.697417
7.555556
69.79167
0.00209248
50.54303
3.05697
0.4973239
	
185
0184.00
27.03151
3.251772
27.54491
40.78947
1.815184
43.75
39.32927
0.001673101
35.20933
-8.177823
-1.297175
	
292
0286.00
43.01288
3.711843
21.9697
29.17706
3.732667
27.04918
43.54839
0.001429056
37.43548
5.5774
0.8872145
	
334
0325.02
26.84015
3.508565
22.28261
33.10345
2.585823
28.57143
41.92593
0.0009907295
33.5519
-6.711754
-1.064183
	
202
0197.00
28.34994
2.40906
24.03433
48.54167
1.994467
42.85714
45.61028
8.764326e-05
30.02045
-1.670502
-0.2652086

In [17]:

    

# Tableau de données sans les valeurs aberrantes (Cook > 8 / n )
dataSansOutliers <- a[a$cook  < 8/nobs, ]
dataSansOutliers$cook <- NULL
dataSansOutliers$ypredit <- NULL
dataSansOutliers$res <- NULL
dataSansOutliers$res_std <- NULL
head(dataSansOutliers)


    

    
Out[17]:





srnom
FAIBREVPCT
SqrtChom
MONOPCT
menage1per
SqrtImmig
pasecol1524
tpspartiel

	
59
0061.00
30.95238
3.216338
36.84211
53.84615
1.856953
15.38462
45.90164
	
52
0053.00
55.03356
3.576408
13.04348
68.57143
3.172206
48.27586
59.74026
	
83
0085.00
44.28571
2.995723
32.35294
29.82456
1.43839
37.93103
32.25806
	
151
0150.00
33.69565
4.082483
26.04167
53.06859
2.464498
56.25
49.52077
	
42
0043.00
42.85714
3.867897
11.76471
62.10526
1.301889
17.3913
50.80645
	
50
0051.00
44.61538
2.773501
16.27907
57.92683
2.438843
19.60784
62.43094

In [18]:

    

# Nouveau modèle de régression sans les valeurs aberrantes
ols2 <- lm(FAIBREVPCT ~ SqrtChom+MONOPCT+menage1per+SqrtImmig+pasecol1524+tpspartiel, data = dataSansOutliers)
summary(ols2)
opar <- par(mfrow = c(2, 2), oma = c(0, 0, 1.1, 0))
plot(ols, las = 1)
par(opar)


    

    
Out[18]:





Call:
lm(formula = FAIBREVPCT ~ SqrtChom + MONOPCT + menage1per + SqrtImmig + 
    pasecol1524 + tpspartiel, data = dataSansOutliers)

Residuals:
     Min       1Q   Median       3Q      Max 
-15.4311  -3.7694  -0.2836   3.3556  14.3676 

Coefficients:
             Estimate Std. Error t value Pr(>|t|)    
(Intercept) -38.07012    1.98689 -19.161  < 2e-16 ***
SqrtChom      8.62818    0.56045  15.395  < 2e-16 ***
MONOPCT       0.62153    0.04414  14.082  < 2e-16 ***
menage1per    0.13012    0.02321   5.605 3.52e-08 ***
SqrtImmig     2.37947    0.34582   6.881 1.87e-11 ***
pasecol1524   0.13829    0.02593   5.333 1.49e-07 ***
tpspartiel    0.30772    0.04772   6.448 2.77e-10 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 5.404 on 480 degrees of freedom
Multiple R-squared:  0.842,	Adjusted R-squared:  0.8401 
F-statistic: 426.5 on 6 and 480 DF,  p-value: < 2.2e-16