How to split a CHR column, pivot, then combine tables? - r

Question

So I have two tables:
LST data (24 months in total) (already pivoted_longer)
Buffer Date LST
<chr> <chr> <chr>
1 100 15/01/2010 6.091741043
2 100 16/02/2010 6.405879111
3 100 20/03/2010 8.925945159
4 100 24/04/2011 6.278147269
5 100 07/05/2010 6.133940129
6 100 08/06/2010 7.705591939
7 100 13/07/2011 4.066052173
8 100 11/08/2010 5.962087092
9 100 12/09/2010 5.761892842
10 100 17/10/2011 3.155769317
# ... with 1,550 more rows
Weather data (24 months in total)
Weather variable 15/01/2010 16/02/2010 20/03/2010 24/04/2011 07/05/2010
1 Temperature 12.0 15.0 16.0 23.00 21.50
2 Wind_speed 10.0 9.0 10.5 19.50 9.50
3 Wind_trend 1.0 1.0 1.0 0.00 1.00
4 Wind_direction 22.5 45.0 67.5 191.25 56.25
5 Humidity 40.0 44.5 22.0 24.50 7.00
6 Pressure 1024.0 1018.5 1025.0 1005.50 1015.50
7 Pressure_trend 1.0 1.0 1.0 1.00 1.00
If I pivot the weather data I get:
1 Temperature 15/01/2010 12
2 Temperature 16/02/2010 15
3 Temperature 20/03/2010 16
4 Temperature 24/04/2011 23
5 Temperature 07/05/2010 21.5
6 Temperature 08/06/2010 36.5
7 Temperature 13/07/2011 33
8 Temperature 11/08/2010 34.5
9 Temperature 12/09/2010 33
10 Temperature 17/10/2011 27
# ... with 158 more rows
(each weather variable listed in turn).
I need to combine 1) and 3) - using the date and something like data_long <- merge(LST_data,weather_data,by="Date") I think - appending weather data columns to each row in 1).
But I'm stuck.

Answer 1

The solution I found to this was to pivot the weather data (longer):
weather_long <- weather %>% pivot_longer(cols = 2:21, names_to = "Date", values_to = "Value")
which gives a tibble in the format:
# A tibble: 180 x 3
`Weather variable` Date Value
<chr> <chr> <dbl>
1 Temperature 28/10/2016 17
2 Temperature 31/12/2016 22
3 Temperature 16/01/2017 25
4 Temperature 05/03/2017 19
(as described above in the question).
Because this process changes the 'Date' variable type:
tibble [180 x 3] (S3: tbl_df/tbl/data.frame)
$ Weather variable: chr [1:180] "Temperature" "Temperature" "Temperature" "Temperature" ...
$ Date : chr [1:180] "28/10/2016" "31/12/2016" "16/01/2017" "05/03/2017" ...
$ Value : num [1:180] 17 22 25 19 20 22 11 10 3 9 ...
I then corrected this:
weather_long$Date <- as.Date(weather_long$Date, format = "%d/%m/%Y")
Next was to convert the weather data to the 'wide' format (in preparation for the next step):
weather_wide <- weather_long %>%
pivot_wider(names_from = "Weather variable", values_from = "Value")
Then join it to the LST data using the Date column as the key:
LST_Weather_dataset <- full_join(data_long, weather_wide, by = "Date")
This produced the desired result:
str(LST_Weather_dataset)
'data.frame': 380 obs. of 16 variables:
$ Buffer : int 100 200 300 400 500 600 700 800 900 1000 ...
$ Date : Date, format: "2016-10-28" "2016-10-28" "2016-10-28" "2016-10-28" ...
$ LST : num 0.918 0.951 0.791 0.748 0.687 ...
$ Month : num 10 10 10 10 10 10 10 10 10 10 ...
$ Year : num 2016 2016 2016 2016 2016 ...
$ JulianDay : num 302 302 302 302 302 302 302 302 302 302 ...
$ TimePeriod : Factor w/ 2 levels "1","2": 1 1 1 1 1 1 1 1 1 1 ...
$ Temperature : num 17 17 17 17 17 17 17 17 17 17 ...
$ Humidity : num 59 59 59 59 59 59 59 59 59 59 ...
$ Humidity_trend: num 1 1 1 1 1 1 1 1 1 1 ...
$ Wind_speed : num 19 19 19 19 19 19 19 19 19 19 ...
$ Wind_gust : num 0 0 0 0 0 0 0 0 0 0 ...
$ Wind_trend : num 2 2 2 2 2 2 2 2 2 2 ...
$ Wind_direction: num 338 338 338 338 338 ...
$ Pressure : num 1017 1017 1017 1017 1017 ...
$ Pressure_trend: num 2 2 2 2 2 2 2 2 2 2 ...