此筆資料是政府開放資料中，桃園市嫌疑犯人數與不同性別、不同教育程度的統計資料

library(readxl)
## Warning: package 'readxl' was built under R version 3.3.3
library(dplyr)
## Warning: package 'dplyr' was built under R version 3.3.3
## 
## Attaching package: 'dplyr'
## The following objects are masked from 'package:stats':
## 
##     filter, lag
## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union
library(ggplot2)
## Warning: package 'ggplot2' was built under R version 3.3.3
library(magrittr)
## Warning: package 'magrittr' was built under R version 3.3.3
#匯入資料
dta <-  read.csv(file = "m_edu.csv", header = TRUE)
#整理資料，把嫌疑犯人數都換成數值資料
dta$count <- as.numeric(dta$count)
#看資料基本統計
summary(dta)
##       year         e_clasee      sex          count      
##  Min.   : 89   大專    :45   女    :120   Min.   :  1.0  
##  1st Qu.: 92   不識字  :45   全性別:120   1st Qu.: 63.0  
##  Median : 96   自修    :45   男    :120   Median :119.0  
##  Mean   : 96   其他    :45                Mean   :121.9  
##  3rd Qu.:100   研究所  :45                3rd Qu.:180.2  
##  Max.   :103   高中(職):45                Max.   :251.0  
##                (Other) :90

想看看此筆資料性別對嫌疑犯人數的影響；以及最高教育程度與嫌疑犯人數的關係

#載進 ggplot2 準備畫圖
require(ggplot2)
#底下的圖都用黑白配色（theme_bw）
old <- theme_set(theme_bw())
#看不同性別的盒鬚圖
ggplot(data = dta, aes(x = sex, y = count)) +
  geom_boxplot() + coord_flip() +
  labs( y = 'count', x = 'sex', 
        title = 'Taoyuan City suspects and sex relationship')

若只以嫌疑犯人數來看，則發現男性的嫌疑犯人數高於女生

#只看男女兩性別
dta1 <-  dta[-which(dta$sex=="全性別"),]
#檢驗兩組樣本的母體變異數
#H0:兩組樣本的母體變異數相等
var.test(dta1$count[dta1$sex=="女"],dta1$count[dta1$sex=="男"])
## 
##  F test to compare two variances
## 
## data:  dta1$count[dta1$sex == "女"] and dta1$count[dta1$sex == "男"]
## F = 1.0917, num df = 119, denom df = 119, p-value = 0.6329
## alternative hypothesis: true ratio of variances is not equal to 1
## 95 percent confidence interval:
##  0.7608621 1.5664777
## sample estimates:
## ratio of variances 
##            1.09173

p-value=0.6329>0.05故無法拒絕虛無假說，因此假設兩組樣本的母體變異數相等

利用t檢定分析

#以t檢定比較不同性別的嫌疑犯數量差異
#變異數同值下的 t 檢定
#H0:男生的嫌疑犯人數比女生多
t.test(dta1$count~factor(dta1$sex))
## 
##  Welch Two Sample t-test
## 
## data:  dta1$count by factor(dta1$sex)
## t = -1.6744, df = 237.54, p-value = 0.09537
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -33.029342   2.679342
## sample estimates:
## mean in group 女 mean in group 男 
##         114.6583         129.8333

p-value=0.09537>0.05，無法拒絕虛無假說，男生的嫌疑犯人數確實比女生多

#整理資料
#將不同教育程度者，加總嫌疑犯人數
a <- select(dta1,year,e_clasee,count,sex) %>%
        group_by(e_clasee)
#重新命名欄位名稱
colnames(a) <-c("year","edu","count","gender")

看不同教育程度下的嫌疑犯平均數

tapply(a$count, a$edu, mean)
##      大專    不識字      自修      其他    研究所  高中(職)      國小 
## 137.56667 116.10000 112.20000 125.00000  96.13333 131.13333 101.26667 
##      國中 
## 158.56667

最高教育程度在國中者，平均所算出來的嫌疑犯人數最多

看不同教育的嫌疑犯人數差異，以圖來表示

#不同教育程度
a$edu <- factor(a$edu, levels = c('不識字',
                                            '自修',
                                            '國小',
                                            '國中', 
                                            '高中(職)','大專','研究所','其他'))
library(Hmisc)
## Warning: package 'Hmisc' was built under R version 3.3.3
## Loading required package: lattice
## Loading required package: survival
## Warning: package 'survival' was built under R version 3.3.3
## Loading required package: Formula
## Warning: package 'Formula' was built under R version 3.3.3
## 
## Attaching package: 'Hmisc'
## The following objects are masked from 'package:dplyr':
## 
##     combine, src, summarize
## The following objects are masked from 'package:base':
## 
##     format.pval, round.POSIXt, trunc.POSIXt, units
ggplot(data = a, 
       aes(x = edu, y = count))+
  stat_summary(fun.data = 'mean_cl_boot', size = 1) +
  scale_y_continuous(breaks = seq(90, 200, by = 20))+
  geom_hline(yintercept = mean(a$edu) , 
             linetype = 'dotted') +
  labs(x = '最高教育程度', y = '嫌疑犯人數') +
  coord_flip()
## Warning in mean.default(a$edu): argument is not numeric or logical:
## returning NA
## Warning: Removed 1 rows containing missing values (geom_hline).

情況如同平均數的呈現

#因為教育程度有多種類別，用anova檢定
#H0:最高教育程度不同對嫌疑犯的人數沒有影響
anova.result <- aov( count ~ edu, data = a)
summary(anova.result)
##              Df  Sum Sq Mean Sq F value Pr(>F)  
## edu           7   87035   12434   2.623 0.0126 *
## Residuals   232 1099741    4740                 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

p-value=0.0126<0.05故拒絕虛無假說，最高教育程度不同，是會對嫌疑犯人數有影響的

但是我們應該要看比例!

#全部不分性別的情況
#91年教育程度資料
T91<-read_xlsx("91桃園.xlsx")
as <- select(dta,year,e_clasee,count,sex) %>%
        filter(year=="91") 
## Warning: package 'bindrcpp' was built under R version 3.3.3
d<-group_by(as,e_clasee)%>%
  summarise(sum(count))

#與91年嫌疑犯人數資料做合併
names(d)<-c("edu","count")
d$edu<-as.character(d$edu)
joinnew<-left_join(d,T91,by = c("edu","edu"))
joinnew1<-  joinnew[-which(complete.cases(joinnew)=="FALSE"),]
names(joinnew1)<-c("edu","count","total")
ggplot(joinnew1, aes(x=edu, y=count/total)) + geom_bar(colour = 'red',stat="identity")

#103年教育程度資料
T103<-read_xlsx("桃園103.xlsx")
as <- select(dta,year,e_clasee,count,sex) %>%
  filter(year=="103") 
d<-group_by(as,e_clasee)%>%
  summarise(sum(count))

#與103年嫌疑犯人數資料做合併
names(d)<-c("edu","count")
d$edu<-as.character(d$edu)
joinnew<-left_join(d,T103,by = c("edu","edu"))
joinnew2<-  joinnew[-which(complete.cases(joinnew)=="FALSE"),]
names(joinnew2)<-c("edu","count","total")
p = joinnew2$count/joinnew2$total
ggplot(joinnew2, aes(x=edu, y=p)) + geom_bar(colour = 'red',stat="identity")

#91 女
asf91 <- select(dta,year,e_clasee,count,sex) %>%
  filter(year=="91",sex=="女") 
df1<-group_by(asf91,e_clasee)%>%
  summarise(sum(count))

names(df1)<-c("edu","count")
df1$edu<-as.character(df1$edu)
fjoinnew<-left_join(df1,T103,by = c("edu","edu"))
f2joinnew<-  fjoinnew[-which(complete.cases(fjoinnew)=="FALSE"),]
names(f2joinnew)<-c("edu","count","total")
p = f2joinnew$count/f2joinnew$total
ggplot(f2joinnew, aes(x=edu, y=p)) + geom_bar(stat="identity")

#91 男
asm91 <- select(dta,year,e_clasee,count,sex) %>%
  filter(year=="91",sex=="男") 
dm1<-group_by(asm91,e_clasee)%>%
  summarise(sum(count))

names(dm1)<-c("edu","count")
dm1$edu<-as.character(dm1$edu)
mjoinnew<-left_join(dm1,T103,by = c("edu","edu"))
m2joinnew<-  mjoinnew[-which(complete.cases(mjoinnew)=="FALSE"),]
names(m2joinnew)<-c("edu","count","total")
p = m2joinnew$count/m2joinnew$total
ggplot(m2joinnew, aes(x=edu, y=p)) + geom_bar(colour = 'pink',stat="identity")

#分男女來看
#103 女
as1 <- select(dta,year,e_clasee,count,sex) %>%
  filter(year=="103",sex=="女") 
d1<-group_by(as1,e_clasee)%>%
  summarise(sum(count))

names(d1)<-c("edu","count")
d1$edu<-as.character(d1$edu)
fjoinnew<-left_join(d1,T103,by = c("edu","edu"))
f2joinnew<-  fjoinnew[-which(complete.cases(fjoinnew)=="FALSE"),]
names(f2joinnew)<-c("edu","count","total")
p = f2joinnew$count/f2joinnew$total
ggplot(f2joinnew, aes(x=edu, y=p)) + geom_bar(stat="identity")

#103 男
as2 <- select(dta,year,e_clasee,count,sex) %>%
  filter(year=="103",sex=="男") 
d2<-group_by(as2,e_clasee)%>%
  summarise(sum(count))

names(d2)<-c("edu","count")
d2$edu<-as.character(d1$edu)
mjoinnew<-left_join(d2,T103,by = c("edu","edu"))
m2joinnew<-  mjoinnew[-which(complete.cases(mjoinnew)=="FALSE"),]
names(m2joinnew)<-c("edu","count","total")
p = m2joinnew$count/m2joinnew$total
ggplot(m2joinnew, aes(x=edu, y=p)) + geom_bar(colour = 'pink',stat="identity")

1.結論圖:以比例圖來看早年份的91年與103年均可以發現自修者是在嫌疑犯比例中占比較高的 而細看男女的圖: 則發現91年男女均是以自修和不識字的比例比較高，而男生的比例則高於女生

到103男女的比較圖: 女生在103年以自修和不識字的比例比較高 而男生則是不識字的比例最高，其次為國小與研究所，相較於91年，自修比例降低，而研究所與國小的比例稍微上升

但是整體來說，103年的嫌疑犯比例都有比91年的比例下降

2.此資料因為影響的項目為類別資料(性別、教育程度)，受限而沒有使用相關係數與回歸來進一步分析，若未來有加入其他數值資料，則可探討在相同的教育程度下或相同性別下，另一數值變項每增加一單位對嫌疑犯人數的影響，便能掌握什麼變項影響最大