[R] 전처리 샘플 airquality

# airquality NA outlier 처리

# 1. NA outlier 확인

# NA - Ozone Solar.R
# 이상치 - Ozone Wind 
summary(airquality)
sum(is.na(airquality))
boxplot(airquality)

 

# 2. NA 처리

median(airquality$Ozone)  # NA ??????? 31.5
fivenum(airquality$Ozone)
median(airquality$Solar.R)  # NA ????? 205.0

airquality.clean <- airquality
airquality.clean$Ozone <- ifelse(is.na(airquality.clean$Ozone),
                                 31.5,
                                 airquality.clean$Ozone)
airquality.clean$Solar.R <- ifelse(is.na(airquality.clean$Solar.R),
                                   205.0,
                                 airquality.clean$Solar.R)

summary(airquality.clean)
sum(is.na(airquality.clean))

 

# 3. outlier 처리 : 최대값으로 대체

boxplot(airquality.clean) # 중위수 대체를 하니 Ozone outlier가 많이 증가
(Ozone.q1 <- fivenum(airquality.clean$Ozone)[2])
(Ozone.q3 <- fivenum(airquality.clean$Ozone)[4])
Ozone.iqr <- Ozone.q3 - Ozone.q1
Ozone.min <- Ozone.q1 - 1.5* Ozone.iqr
Ozone.max <- Ozone.q3 + 1.5* Ozone.iqr

airquality.clean$Ozone <- ifelse(airquality.clean$Ozone > Ozone.max,
                                 Ozone.max,
                                 airquality.clean$Ozone)

(Wind.q1 <- fivenum(airquality.clean$Wind)[2])
(Wind.q3 <- fivenum(airquality.clean$Wind)[4])
Wind.iqr <- Wind.q3 - Wind.q1
Wind.min <- Wind.q1 - 1.5* Wind.iqr
Wind.max <- Wind.q3 + 1.5* Wind.iqr

airquality.clean$Wind <- ifelse(airquality.clean$Wind > Wind.max,
                                 Wind.max,
                                 airquality.clean$Wind)
boxplot(airquality.clean)

 

# 4. 결과 확인

par(mfrow=c(1,2))
hist(airquality$Ozone,xlim=c(0,200),ylim=c(0,50),breaks = 12, main="airquality")
hist(airquality.clean$Ozone,xlim=c(0,200),ylim=c(0,50),breaks = 6, main="preprocessed")

 

가운데는 왜 튀어나온것일까요?

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

NA  를 중위수로 대체했었지요.