데이터마이너를 꿈꾸며

> library(httr)
> library(stringr)
> ad.data<-read.csv("./ad_result.csv",header = T, stringsAsFactors = F)
> ad.data #tvcm: tv광고금액, magazine: 잡지광고금액,install=신규고객수
     month  tvcm magazine install
1  2013-01  6358     5955   53948
2  2013-02  8176     6069   57300
3  2013-03  6853     5862   52057
4  2013-04  5271     5247   44044
5  2013-05  6473     6365   54063
6  2013-06  7682     6555   58097
7  2013-07  5666     5546   47407
8  2013-08  6659     6066   53333
9  2013-09  6066     5646   49918
10 2013-10 10090     6545   59963
#TV 광고와 신규 고객 산점도 
> ggplot(ad.data,aes(x=tvcm,y=install))+geom_point()+xlab('TV 광고비')+ylab('신규 유저수')+
+ scale_x_continuous(label=comma)+scale_y_continuous(label=comma)

#잡지 광고와 신규 고객 산점도
> ggplot(ad.data,aes(x=magazine,y=install))+geom_point()+xlab('잡지 광고비')+ylab('신규 유저수')+
+ scale_x_continuous(label=comma)+scale_y_continuous(label=comma)

#회귀 분석
> fit<-lm(install~.,data=ad.data[,c("install","tvcm","magazine")])
> fit

Call:
lm(formula = install ~ ., data = ad.data[, c("install", "tvcm", 
    "magazine")])

Coefficients:#모델식
(Intercept)         tvcm     magazine  
    188.174        1.361        7.250  
# 신규 고객=188.174 + TV광고금액x1.361 + 잡지광고금액x7.25

#회귀 분석 요약
> summary(fit)

Call:
lm(formula = install ~ ., data = ad.data[, c("install", "tvcm", 
    "magazine")])

Residuals:#1Q 절대값이 3Q절대값보다 커서 치우침
     Min       1Q   Median       3Q      Max 
-1406.87  -984.49   -12.11   432.82  1985.84 

Coefficients:
             Estimate Std. Error t value Pr(>|t|)   
(Intercept)  188.1743  7719.1308   0.024  0.98123   
tvcm           1.3609     0.5174   2.630  0.03390 * 
magazine       7.2498     1.6926   4.283  0.00364 **
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Residual standard error: 1387 on 7 degrees of freedom
#결정계수와 조정 결정계수. 1에 가까울 수록 적합함
Multiple R-squared:  0.9379,	Adjusted R-squared:  0.9202 
F-statistic: 52.86 on 2 and 7 DF,  p-value: 5.967e-05

library(httr)

library(rvest)
library(stringr)

for(i in 1:100){
shtml='.shtml'
  if(i==1){
url=sprintf('http://business.sohu.com/jrjg/index%s',shtml)
  }else{
page<-324-i
url=paste0('http://business.sohu.com/jrjg/index_',page,shtml)
  }
}

h<-read_html(url)
article.titles=html_nodes(h,'div.f14list ul li a') #문제점1
article.links=html_attr(article.titles,'href') #문제점1

titles=c()
contents=c()
dates=c()

for(link in article.links){
h.link=read_html(link)

#제목 추출
title=repair_encoding(html_text(html_nodes(h.link,'div.left
h1')),from='UTF-8') #문제점2
titles=c(titles,title)

#내용 추출 
content=repair_encoding(html_text(html_nodes(h.link,'div#contentText')),from='UTF-8')
contents=c(contents,content)

#날짜 추출
date=repair_encoding(html_text(html_nodes(h.link,'div.sourceTime')),from='UTF-8')
date=str_extract(date,'[0-9]+.[0-9]+.[0-9]+.') 
dates=c(dates,date)
}

result=data.frame(titles,contents,dates)
View(result) #문제점3

setwd('d:/temp/r_temp')
write.table(result,'sohu.csv',sep=',')
sohu<-read.csv('sohu.csv')#문제점3
sohu[1,]