R Replace row values to the other rows based on/condition on column information in one data set?

user_5104259 0
Viewed 293

I received a flat data, and the values were missing while flatting data. I have to bring the hours up to NAs in hours, based on id, type and Date, so that remove NAs in dollars 

id<-c(1,2,1,1,1,2,1)
dollar<-as.numeric(c(100,200,300,500, NA, NA,NA))
hours<-as.numeric(c(NA,NA, NA, NA, 5,10,12))
type<-c("Engineer", "Engineer","Operating","Part", "Engineer","Engineer","Operating" )
Date<-c("2020-01-02","2020-01-03","2020-01-02","2020-01-04", "2020-01-02","2020-01-03","2020-01-02")

  id dollar hours      type       Date
1  1    100  <NA>  Engineer 2020-01-02
2  2    200  <NA>  Engineer 2020-01-03
3  1    300  <NA> Operating 2020-01-02
4  1    500  <NA>      Part 2020-01-04
5  1   <NA>     5  Engineer 2020-01-02
6  2   <NA>    10  Engineer 2020-01-03
7  1   <NA>    12 Operating 2020-01-02

and I would like to reform my data as below.

  id dollar hours      type       Date
1   1     100      5  Engineer 2020-01-02
2   2     200     10  Engineer 2020-01-03
3   1     300     12 Operating 2020-01-02
4   1     500   <NA>      Part 2020-01-04

It is not just grouped by id, but matches with type and date. 'id' has categories, 'type' has 17 categories and 'Date' are 3 years.

Please help me on this.

conditional-statements grouping r
3 Answers

Here is a dplyr option using summarise

library(dplyr)
df %>%
    group_by(id, type, Date) %>%
    summarise_at(vars(dollar, hours), ~mean(.x, na.rm = T))
## A tibble: 4 x 5
## Groups:   id, type [4]
#     id type      Date       dollar hours
#  <dbl> <fct>     <fct>       <dbl> <dbl>
#1     1 Engineer  2020-01-02    100     5
#2     1 Operating 2020-01-02    300    12
#3     1 Part      2020-01-04    500   NaN
#4     2 Engineer  2020-01-03    200    10

Or even 

df %>% group_by(id, type, Date) %>% summarise_all(~mean(.x, na.rm = T))

Sample data

df <- data.frame(id, dollar, hours, type, Date)
user_65309700

Here is one approach with tidyverse. You can group by id, type, and date and then fill missing NA with available values.

library(tidyverse)

df %>%
  group_by(id, type, Date) %>%
  fill(c(dollar, hours), .direction = "updown") %>%
  slice(1)

Output

# A tibble: 4 x 5
# Groups:   id, type, Date [4]
     id dollar hours type      Date      
  <dbl>  <dbl> <dbl> <fct>     <fct>     
1     1    100     5 Engineer  2020-01-02
2     1    300    12 Operating 2020-01-02
3     1    500    NA Part      2020-01-04
4     2    200    10 Engineer  2020-01-03
user_34606700

Alternatively, you can use pivot_longer and pivot_wider function from tidyr to reshape your dataset into a longer format, remove NA values and then reshape into a wider format:

library(dplyr)
library(tidyr)
DF %>% pivot_longer(cols = c(dollar, hours), names_to = "var", values_to = "val") %>%
  filter(!is.na(val)) %>% pivot_wider(names_from = var, values_from = val)

# A tibble: 4 x 5
     id type      Date       dollar hours
  <dbl> <fct>     <fct>       <dbl> <dbl>
1     1 Engineer  2020-01-02    100     5
2     2 Engineer  2020-01-03    200    10
3     1 Operating 2020-01-02    300    12
4     1 Part      2020-01-04    500    NA
user_122598530
Related