Introduction to Data Frames in R

Student Statistics at National Tsing Hua University

# load libraries
library(readr)
library(dplyr)

# load data frame
Students <- read_csv('total_student_sex.csv')

Statistics of registered students over the years (in total)

# inspect data frame
Students

# inspect 6 rows
head(Students)

# inspect 10 rows
head(Students, 10)

summary(Students)

 #Student Statistics at National Tsing Hua University#\r\nYear   Semester         (A)Undergraduate_Male
 Length:141                                                    Length:141         Length:141           
 Class :character                                              Class :character   Class :character     
 Mode  :character                                              Mode  :character   Mode  :character     
 (A)Undergraduate_Female (A)Undergraduate_Count (B)Master_Male     (B)Master_Female   (B)Master_Count   
 Length:141              Length:141             Length:141         Length:141         Length:141        
 Class :character        Class :character       Class :character   Class :character   Class :character  
 Mode  :character        Mode  :character       Mode  :character   Mode  :character   Mode  :character  
 (C)Doctor_Male     (C)Doctor_Female   (C)Doctor_Count    Total(A+B+C)\r\n_Male Total(A+B+C)_Female
 Length:141         Length:141         Length:141         Length:141            Length:141         
 Class :character   Class :character   Class :character   Class :character      Class :character   
 Mode  :character   Mode  :character   Mode  :character   Mode  :character      Mode  :character   
 Total(A+B+C)_Count male/female(ratio) (B+C)\r\nGraduate_Male (B+C)Graduate_Female (B+C)\r\nGraduate_Count
 Length:141         Length:141         Length:141             Length:141           Length:141             
 Class :character   Class :character   Class :character       Class :character     Class :character       
 Mode  :character   Mode  :character   Mode  :character       Mode  :character     Mode  :character       
      R                S              T                U              ...23        
 Length:141         Mode:logical   Mode:logical   Length:141         Mode:logical  
 Class :character   NA's:141       NA's:141       Class :character   NA's:141      
 Mode  :character                                 Mode  :character                 

# inspect data frame with pipe
Students %>%
  head()

# select one column
Student_Undergraduate_Count <- Students %>% 
  select("(A)Undergraduate_Count")
Student_Undergraduate_Count

# select multiple columns
Student_Undergraduate <- Students %>%
  select(Semester,"(A)Undergraduate_Male","(A)Undergraduate_Female","(A)Undergraduate_Count")

Student_Undergraduate

# select all columns except one
no_ratio <- Students %>%
  select(-"male/female(ratio)")
no_ratio

# select all columns except a set
Student_total <- Students %>%
  select(-"(A)Undergraduate_Male",-"(A)Undergraduate_Female", -"(B)Master_Male", -"(B)Master_Female", -"(C)Doctor_Male", -"(C)Doctor_Female")
Student_total

# filter rows one condition
total_fall <- Students %>%
  filter(Semester == 'Fall semester(Total)')
total_fall

# filter rows multiple conditions
total_male_more_female <- Students %>%
  filter(Semester == 'Fall semester(Nanda campus)', as.numeric(`male/female(ratio)`) < 1 )
total_male_more_female

Undergraduate (male > female) or total (male > female)

# filter rows with or
FallMain_and_ratio_less1 <- Students %>%
  filter(as.numeric(`(A)Undergraduate_Female`) > as.numeric(`(A)Undergraduate_Male`) | as.numeric(`male/female(ratio)`) < 1)

FallMain_and_ratio_less1

NA