Styjun

I have one data frame like:

 a b c d e f g 
 1 Car 10/02 01/02 30/02 14 1 NA
 2 Car 10/02 07/02 20/02 0 NA 7

I want to get :

 a b c d e f g 
 1 Car 10/02 01/02 20/02 14 1 7

Like a group by (a,b) , select the min Date for c and d, select the max for e and select the non-null for f and g

How can I solve it in R ?

Using dplyr we can group_by a and b, convert c and d to actual dates using dmy from lubridate and select the minimum date, select maximum value of e and non-NA value of f and g.

library(dplyr)
library(lubridate)

df %>%
 group_by(a, b) %>%
 summarise(c = c[which.min(dmy(paste0(c, "/19")))], 
 d = d[which.min(dmy(paste0(d, "/19")))], 
 e = max(e), 
 f = f[!is.na(f)], 
 g = g[!is.na(g)])

# a b c d e f g
# <fct> <fct> <fct> <fct> <dbl> <int> <int>
#1 Car 10/02 01/02 20/02 14 1 7

As shown in the example, I am assuming you would have only one non-NA value for f and g if you have more than one then use which.max to select the first non-NA value from those columns.

Use library dplyr, let's say your dataframe is df. First thing is replace NA by 0

df[is.na(df)] <- 0 # you can now directly take sum of f and g column in group by
library(dplyr)
df_1 <- df %>% group_by(a,b) %>% summarise(c = min(c),d = min(d),e = max(e),f = sum(f),g = sum(g))

using data.table. I had to change one of your dates because there is no feb 30th

 x <- data.frame( 
 a= c( "Car","Car"), b=c("10/20","10/20"), c=c("01/02","07/02"),d=c("28/02","20/02"), e=c(14,0), f=c(1, NA), g=c(NA,7)) 

 library( data.table) # you may need to install
 x <- data.table( x )

#convert to dates
 x$c <- as.Date(x$c, "%d/%m")
 x$d <- as.Date(x$d, "%d/%m")

# group as you specfied
 x[ ,

 .(
 c = min( c ),
 d = min( d ),
 e = max( e ),
 f= max( f , na.rm=T),
 g= max( g , na.rm=T)
 ) ,

 by= c( "a","b")
 ]