[R] Scatter plot with Regression lines (ggplot2)

 

• 데이터 집계 및 피벗

 # count Patients
 a <- data[,c(3,39,43)] %>% 
   group_by(age_group2, year) %>% 
   summarise(n_patient = n()) 
 
 # pivot wider
 b <- a%>% 
   dcast(year ~ ..., value.var = 'n_patient')

  patient_id  year age_group2
  <chr>      <int> <fct>     
1 0005969     2014 61-75     
2 0010250     2019 >=76      
3 0013541     2011 >=76      
4 0013600     2011 46-60     
5 0024285     2016 46-60     
6 0025533     2018 61-75     

>   head(a)
# A tibble: 6 × 3
# Groups:   age_group2 [1]
  age_group2  year n_patient
  <fct>      <int>     <int>
1 <31         2011         1
2 <31         2012         1
3 <31         2013         4
4 <31         2015         4
5 <31         2016         4
6 <31         2017         4

>   head(b)
  year <31 31-45 46-60 61-75 >=76
1 2010  NA     2    10     5    4
2 2011   1     7    12    10    5
3 2012   1     4     8    17   14
4 2013   4    10    15    25   17
5 2014  NA    11    24    23    7
6 2015   4    14    17    24    9

 

 

• 회귀선을 포함한 산점도 

# scatter plot with regression lines
 ggplot(data=a, aes(y = a$n_patient, x= a$year, color=age_group2)) +
   geom_point() +
   geom_smooth(method = 'lm', se=F)+
   labs(title = 'Patients per year, stratified for age',
        x = "Year",
        y = "Patients")