1.2 lnitial data analysis Solution:
melbdata <- read.csv("Melbourne_housing_FULL.csv")
library(dplyr)
##
## 载入程辑包:'dplyr'
## The following objects are masked from 'package:stats':
##
## filter, lag
## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, union
melbdata.sub <- melbdata %>%
filter(Suburb %in% c("Hawthorn", "Brunswick", "Craigieburn")) %>%
mutate(Suburb = factor(Suburb, levels = c("Craigieburn", "Brunswick", "Hawthorn")))
1.3 Finding associationl Solution:
lm1 <- lm(data = melbdata.sub, Price/1000 ~ BuildingArea)
library(ggplot2)
library(dplyr)
library(ggfortify)
melbdata.sub_out <- melbdata %>%
filter(Suburb %in% c("Hawthorn", "Brunswick", "Craigieburn")) %>%
mutate(Suburb = factor(Suburb, levels = c("Craigieburn", "Brunswick","Hawthorn"))) %>%
slice(-c(158,682))
1.4 Finding association Ⅰ Solution:
lm2 <- lm(Price ~ Suburb + BuildingArea, data = melbdata.sub)
summary(lm2)
##
## Call:
## lm(formula = Price ~ Suburb + BuildingArea, data = melbdata.sub)
##
## Residuals:
## Min 1Q Median 3Q Max
## -4074050 -248461 -26674 166018 5479862
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -150520.6 78836.2 -1.909 0.0569 .
## SuburbBrunswick 660277.3 72911.5 9.056 <2e-16 ***
## SuburbHawthorn 1060993.2 79615.5 13.326 <2e-16 ***
## BuildingArea 4435.4 346.9 12.786 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 609900 on 414 degrees of freedom
## (因为不存在,709个观察量被删除了)
## Multiple R-squared: 0.425, Adjusted R-squared: 0.4208
## F-statistic: 102 on 3 and 414 DF, p-value: < 2.2e-16
boxplot(Price ~ Suburb, data=melbdata.sub)
SuburbBrunswick has a coefficient of 781107.7, which means that home values in SuburbBrunswick are on average $781107.7 higher when compared to other suburbs. SuburbHawthorn has a coefficient of 1144544.5, which means that home values in SuburbHawthorn are on average $1144544.5 higher when compared to other suburbs.
Price=
melbdata.sub$Suburb=as.factor(melbdata.sub$Suburb)
my.colors <- rgb(c(1,0,0),c(0,1,0),c(0,0,1),alpha = 0.5)
plot(Price ~ BuildingArea, data = melbdata.sub_out,
main = "House price of some suburbs against Building Area",
xlab = "Building Area(in sqm)",
ylab = "Price ($AUD)",
col = my.colors[as.integer(melbdata.sub_out[["Suburb"]])],
pch = 19)
legend("topright", legend = levels(melbdata.sub_out[["Suburb"]]),
col=my.colors, pch = 19)
obs_sub <- subset(melbdata.sub_out, select = c("BuildingArea", "Suburb"))
buArea_sub <- with(obs_sub, split(BuildingArea, Suburb))
pred1 = data.frame(BuildingArea = buArea_sub[["Brunswick"]], Suburb = "Brunswick")
lines(pred1[["BuildingArea"]], predict(lm2, newdata = pred1),col='green')
pred2 = data.frame(BuildingArea = buArea_sub[["Craigieburn"]], Suburb = "Craigieburn")
lines(pred2[["BuildingArea"]], predict(lm2, newdata = pred2),col='red')
pred3 = data.frame(BuildingArea = buArea_sub[["Hawthorn"]], Suburb = "Hawthorn")
lines(pred3[["BuildingArea"]], predict(lm2, newdata = pred3),col='blue')
library(ggfortify)
autoplot(lm2, which=1:6, nrow=3, ncol=2)
lm3 <-lm(Price ~ Suburb + BuildingArea+ Car,data=melbdata.sub)
summary(lm3)
##
## Call:
## lm(formula = Price ~ Suburb + BuildingArea + Car, data = melbdata.sub)
##
## Residuals:
## Min 1Q Median 3Q Max
## -3417281 -273352 -59191 252474 5001704
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -447547.1 89421.2 -5.005 8.37e-07 ***
## SuburbBrunswick 781107.7 73776.4 10.588 < 2e-16 ***
## SuburbHawthorn 1144544.5 78263.7 14.624 < 2e-16 ***
## BuildingArea 3761.7 351.3 10.708 < 2e-16 ***
## Car 220740.5 34616.8 6.377 4.98e-10 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 588500 on 402 degrees of freedom
## (因为不存在,720个观察量被删除了)
## Multiple R-squared: 0.477, Adjusted R-squared: 0.4718
## F-statistic: 91.66 on 4 and 402 DF, p-value: < 2.2e-16
autoplot(lm3, which=1:6, nrow=3, ncol=2)