전처리 Part.4

전처리 Part.4에서는 R을 사용해 연도별 날씨 데이터를 불러와서 전처리를 진행했다.

#패키지 불러오기
library(dplyr)
library(tidyr)
library(stringr)
library(sp)
library(rgdal)
library(rgeos)
library(tmap)
library(raster)
library(spdep)
library(gstat)
library(spgwr)
library(GWmodel)

전처리 Part.4(by using R)

연도별 weather 데이터 불러오기

세계기상기구(WMO)는 기온, 강수량 등의 기후요소가 평년값에 비해 현저히 높거나 낮은 수치(90퍼센타일 또는 10퍼센타일 미만의 범위)를 이상 기후로 정의하고 있다.

이 기준에 따라서 2017 ~ 2019년의 weather 데이터의 이상치를 분류하도록 한다.

• 기온과 풍속, 습도, 지면 온도의 데이터 전처리: 1년 별로 비교해서 이상치(하위 10% 미만, 상위 10% 초과) 안에 포함되는 경우 1, 포함되지 않는 경우에는 0

• 강수량, 적설량 데이터 전처리: 당해 연도의 상위 10% 백분위 수에 포함되는 경우 1, 포함되지 않는 경우에는 0

• 안개 데이터 전처리: 안개는 지속 시간이 조금이라도 있으면 1, 없으면 0

1) 2017년의 weather 데이터 처리

weather_2017 <- read.csv('weather_2017.csv')

temp_average <- c() # 평균온도
qt <- quantile(weather_2017[, names(weather_2017)[3]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
temp_average <- ifelse(weather_2017[, names(weather_2017)[3]] < qt[2] | weather_2017[, names(weather_2017)[3]] > qt[3], 1, 0)
# 10퍼센타일 미만이거나 90퍼센타일 이상이면 1 아니면 0

#나머지도 동일하게 적용
temp_min <- c() #최저온도
qt <- quantile(weather_2017[, names(weather_2017)[4]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
temp_min <- ifelse(weather_2017[, names(weather_2017)[4]] < qt[2] | weather_2017[, names(weather_2017)[4]] > qt[3], 1, 0)

temp_max <- c() #최고온도
qt <- quantile(weather_2017[, names(weather_2017)[5]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
temp_max <- ifelse(weather_2017[, names(weather_2017)[5]] < qt[2] | weather_2017[, names(weather_2017)[5]] > qt[3], 1, 0)

wind_max <- c() #최대풍속
qt <- quantile(weather_2017[, names(weather_2017)[7]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
wind_max <- ifelse(weather_2017[, names(weather_2017)[7]] < qt[2] | weather_2017[, names(weather_2017)[7]] > qt[3], 1, 0)

wind_average <- c() #평균풍속
qt <- quantile(weather_2017[, names(weather_2017)[9]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
wind_average <- ifelse(weather_2017[, names(weather_2017)[9]] < qt[2] | weather_2017[, names(weather_2017)[9]] > qt[3], 1, 0)

humidity_average <- c() #평균습도
qt <- quantile(weather_2017[, names(weather_2017)[10]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
humidity_average <- ifelse(weather_2017[, names(weather_2017)[10]] < qt[2] | weather_2017[, names(weather_2017)[10]] > qt[3], 1, 0)

road_average <- c() #평균지면온도
qt <- quantile(weather_2017[, names(weather_2017)[12]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
road_average <- ifelse(weather_2017[, names(weather_2017)[12]] < qt[2] | weather_2017[, names(weather_2017)[12]] > qt[3], 1, 0)

weather_2017$평균온도 <- temp_average
weather_2017$최저온도 <- temp_min
weather_2017$최고온도 <- temp_max
weather_2017$최대풍속 <- wind_max
weather_2017$평균풍속 <- wind_average
weather_2017$평균습도 <- humidity_average
weather_2017$평균지면온도 <- road_average

# NA 값은 이상치가 없는 경우이므로 0으로 처리
weather_2017[, names(weather_2017)[6]] <- ifelse(is.na(weather_2017[, names(weather_2017)[6]]), 0, weather_2017[, 6])
weather_2017[, names(weather_2017)[11]] <- ifelse(is.na(weather_2017[, names(weather_2017)[11]]), 0, weather_2017[, names(weather_2017)[11]])

day_rain <- c()
qt <- quantile(weather_2017[, names(weather_2017)[6]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
day_rain <- ifelse(weather_2017[, names(weather_2017)[6]] > qt[3], 1, 0)

day_snow <- c()
qt <- quantile(weather_2017[, names(weather_2017)[11]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
day_snow <- ifelse(weather_2017[, names(weather_2017)[11]] > qt[3], 1, 0)

weather_2017$강수량 <- day_rain
weather_2017$적설량 <- day_snow
weather_2017$안개시간 <- ifelse( is.na(weather_2017[, names(weather_2017)[13]]), 0, 1)

weather_2017_arrange <- weather_2017[,c(2,14:ncol(weather_2017))]  #새로 만든 변수 확인
dim(weather_2017_arrange)
weather_2017_arrange %>% head()

2) 2018년도 마찬가지

weather_2018 <- read.csv('weather_2018.csv')

temp_average <- c()
qt <- quantile(weather_2018[, names(weather_2018)[3]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
temp_average <- ifelse(weather_2018[, names(weather_2018)[3]] < qt[2] | weather_2018[, names(weather_2018)[3]] > qt[3], 1, 0)

temp_min <- c()
qt <- quantile(weather_2018[, names(weather_2018)[4]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
temp_min <- ifelse(weather_2018[, names(weather_2018)[4]] < qt[2] | weather_2018[, names(weather_2018)[4]] > qt[3], 1, 0)

temp_max <- c()
qt <- quantile(weather_2018[, names(weather_2018)[5]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
temp_max <- ifelse(weather_2018[, names(weather_2018)[5]] < qt[2] | weather_2018[, names(weather_2018)[5]] > qt[3], 1, 0)

wind_max <- c()
qt <- quantile(weather_2018[, names(weather_2018)[7]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
wind_max <- ifelse(weather_2018[, names(weather_2018)[7]] < qt[2] | weather_2018[, names(weather_2018)[7]] > qt[3], 1, 0)

wind_average <- c()
qt <- quantile(weather_2018[, names(weather_2018)[9]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
wind_average <- ifelse(weather_2018[, names(weather_2018)[9]] < qt[2] | weather_2018[, names(weather_2018)[9]] > qt[3], 1, 0)

humidity_average <- c()
qt <- quantile(weather_2018[, names(weather_2018)[10]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
humidity_average <- ifelse(weather_2018[, names(weather_2018)[10]] < qt[2] | weather_2018[, names(weather_2018)[10]] > qt[3], 1, 0)

road_average <- c()
qt <- quantile(weather_2018[, names(weather_2018)[12]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
road_average <- ifelse(weather_2018[, names(weather_2018)[12]] < qt[2] | weather_2018[, names(weather_2018)[12]] > qt[3], 1, 0)

weather_2018$평균온도 <- temp_average
weather_2018$최저온도 <- temp_min
weather_2018$최고온도 <- temp_max
weather_2018$최대풍속 <- wind_max
weather_2018$평균풍속 <- wind_average
weather_2018$평균습도 <- humidity_average
weather_2018$평균지면온도 <- road_average

#NA 값 0으로 처리
weather_2018[, names(weather_2018)[6]] <- ifelse(is.na(weather_2018[, names(weather_2018)[6]]), 0, weather_2018[, names(weather_2018)[6]])
weather_2018[, names(weather_2018)[11]] <- ifelse(is.na(weather_2018[, names(weather_2018)[11]]), 0, weather_2018[, names(weather_2018)[11]])

day_rain <- c()
qt <- quantile(weather_2018[, names(weather_2018)[6]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
day_rain <- ifelse(weather_2018[, names(weather_2018)[6]] > qt[3], 1, 0)

day_snow <- c()
qt <- quantile(weather_2018[, names(weather_2018)[11]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
day_snow <- ifelse(weather_2018[, names(weather_2018)[11]] > qt[3], 1, 0)

weather_2018$강수량 <- day_rain
weather_2018$적설량 <- day_snow
weather_2018$안개시간 <- ifelse( is.na(weather_2018[, names(weather_2018)[13]]), 0, 1)

weather_2018_arrange <- weather_2018[,c(2,14:ncol(weather_2018))]
dim(weather_2018_arrange)
weather_2018_arrange %>% head()

3) 2019년 weather 데이터처리

weather_2019 <- read.csv('weather_2019.csv')

qt <- quantile(weather_2019[, names(weather_2019)[3]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
temp_average <- ifelse(weather_2019[, names(weather_2019)[3]] < qt[2] | weather_2019[, names(weather_2019)[3]] > qt[3], 1, 0)

temp_min <- c()
qt <- quantile(weather_2019[, names(weather_2019)[4]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
temp_min <- ifelse(weather_2019[, names(weather_2019)[4]] < qt[2] | weather_2019[, names(weather_2019)[4]] > qt[3], 1, 0)

temp_max <- c()
qt <- quantile(weather_2019[, names(weather_2019)[5]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
temp_max <- ifelse(weather_2019[, names(weather_2019)[5]] < qt[2] | weather_2019[, names(weather_2019)[5]] > qt[3], 1, 0)

wind_max <- c()
qt <- quantile(weather_2019[, names(weather_2019)[7]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
wind_max <- ifelse(weather_2019[, names(weather_2019)[7]] < qt[2] | weather_2019[, names(weather_2019)[7]] > qt[3], 1, 0)

wind_average <- c()
qt <- quantile(weather_2019[, names(weather_2019)[9]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
wind_average <- ifelse(weather_2019[, names(weather_2019)[9]] < qt[2] | weather_2019[, names(weather_2019)[9]] > qt[3], 1, 0)

humidity_average <- c()
qt <- quantile(weather_2019[, names(weather_2019)[10]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
humidity_average <- ifelse(weather_2019[, names(weather_2019)[10]] < qt[2] | weather_2019[, names(weather_2019)[10]] > qt[3], 1, 0)

road_average <- c()
qt <- quantile(weather_2019[, names(weather_2019)[12]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
road_average <- ifelse(weather_2019[, names(weather_2019)[12]] < qt[2] | weather_2019[, names(weather_2019)[12]] > qt[3], 1, 0)

weather_2019$평균온도 <- temp_average
weather_2019$최저온도 <- temp_min
weather_2019$최고온도 <- temp_max
weather_2019$최대풍속 <- wind_max
weather_2019$평균풍속 <- wind_average
weather_2019$평균습도 <- humidity_average
weather_2019$평균지면온도 <- road_average

#NA 값 0으로 처리
weather_2019[, names(weather_2019)[6]] <- ifelse(is.na(weather_2019[, names(weather_2019)[6]]), 0, weather_2019[, names(weather_2019)[6]])
weather_2019[, names(weather_2019)[11]] <- ifelse(is.na(weather_2019[, names(weather_2019)[11]]), 0, weather_2019[, names(weather_2019)[11]])

day_rain <- c()
qt <- quantile(weather_2019[, names(weather_2019)[6]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
day_rain <- ifelse(weather_2019[, names(weather_2019)[6]] > qt[3], 1, 0)

day_snow <- c()
qt <- quantile(weather_2019[, names(weather_2019)[11]], probs=c(0, 0.1, 0.9, 1), na.rm=T)
day_snow <- ifelse(weather_2019[, names(weather_2019)[11]] > qt[3], 1, 0)

weather_2019$강수량 <- day_rain
weather_2019$적설량 <- day_snow
weather_2019$안개시간 <- ifelse( is.na(weather_2019[, names(weather_2019)[13]]), 0, 1)

weather_2019 <- weather_2018[,c(2,14:ncol(weather_2019))]
dim(weather_2019_arrange)
weather_2019_arrange %>% head()

4) 2017 ~ 2019 날씨 데이터 합치기

weather_arrange <- bind_rows(weather_2017_arrange, weather_2018_arrange, weather_2019_arrange)
dim(weather_arrange)
weather_arrange %>% head()

R write.csv(weather_arrange, file='weather_arrange.csv', row.names=F) `