合并R中的行，同时排除某些数据

Question

Let's say I have a data frame with many subjects and many test variables:

   Name      Date1      Date2 `Test1` `Test2` `Test3`
  <dbl>     <dttm>     <dttm>   <chr>   <chr>   <chr>
1 Steve 2012-02-27 2011-11-18    <NA>    <NA>      3
2 Steve 2012-02-27 2012-01-22      4     <NA>    <NA>
3 Steve 2012-02-27 2014-08-09    <NA>      8     <NA>
4 Mike  2012-02-09 2007-03-29      1       2       3
5 Mike  2012-02-09 2009-07-13    <NA>      5       6
6 Mike  2012-02-09 2014-03-11    <NA>    <NA>      9
7 John  2012-03-20 2013-10-22      1       2     <NA>
8 John  2012-03-20 2014-03-17      4       5     <NA>
9 John  2012-03-20 2015-06-01    <NA>      8       9

I would like to know (most likely with dplyr) how to exclude data of rows that have a Date2 that is past Date1. Then to combine the remaining data into one row by (arranged by Name) while excluding the earlier data that have more recent results. Then write a new data frame that excludes the Date2 column, all while still including the "NA"s in the data.
Also, if none of the Date2 column are before the Date1 column, I would like to keep the Name but include a row of "NA"s (as in the case of "John").

So the results should look like this:

   Name      Date1 `Test1` `Test2` `Test3`
  <dbl>     <dttm>   <chr>   <chr>   <chr>
1 Steve 2012-02-27      4     <NA>      3
2 Mike  2012-02-09      1       5       6
3 John  2012-03-20    <NA>    <NA>    <NA>

Any help on this would be greatly appreciated, thank you.

Answer 1

This will do it with dplyr ...

library(dplyr)
df2 <- df %>% filter(as.Date(Date2) <= as.Date(Date1)) %>% #remove date2 past date1
  arrange(as.Date(Date2)) %>% #make sure ordered by date2
  group_by(Name, Date1) %>% #group by name and date1
  summarise_all(function(x) last(x[!is.na(x)])) %>% #summarise remaining (i.e. the test-columns) by the last non-NA value
  right_join(df %>% distinct(Name, Date1)) %>% #join names and date1 from original df (to restore NA rows such as John)
  select(-Date2) #remove Date2

df2

   Name      Date1 Test1 Test2 Test3
1 Steve 2012-02-27     4  <NA>     3
2  Mike 2012-02-09     1     5     6
3  John 2012-03-20  <NA>  <NA>  <NA>