mutate df to generate several new columns using a function on several original fields

问题

I have the following dataframe in R showing several attributes for some community districts (field CD) in two different years:

#Example data with one single attribute

x <- structure(list(numbldgs = c(195, 845, 3621, 3214, 10738, 793, 
223, 957, 4248, 3456, 11576, 803), Year = c("2007", "2007", "2007", 
"2007", "2007", "2007", "2018", "2018", "2018", "2018", "2018", 
"2018"), CD = c("103", "111", "210", "313", "414", "501", "103", 
"111", "210", "313", "414", "501")), row.names = c(NA, -12L), class = c("tbl_df", 
"tbl", "data.frame"))

(thanks to the input of folks here) I have been able to calculate the % increase per CD across time, using:

x2<-x %>% 
arrange(CD) %>% 
group_by(CD) %>% 
mutate(rel_inc= 100*(numbldgs-lag(numbldgs, default=first(numbldgs)))/lag(numbldgs, default=first(numbldgs)))

Now I am trying to apply the same function to several fields of a larger dataset, like the one below.

x<-structure(list(Year = c("2007", "2007", "2007", "2007", "2007", 
"2007", "2018", "2018", "2018", "2018", "2018", "2018"), SFHA_effective = c("yes", 
"yes", "yes", "yes", "yes", "yes", "yes", "yes", "yes", "yes", 
"yes", "yes"), CD = c("103", "111", "210", "313", "414", "501", 
"103", "111", "210", "313", "414", "501"), Abb = c("LES", "EH", 
"CI/Co", "CI", "RA", "SP", "LES", "EH", "CI/Co", "CI", "RA", 
"SP"), numbldgs = c(195, 845, 3621, 3214, 10738, 793, 223, 957, 
4248, 3456, 11576, 803), resunits = c(15174, 18475, 20115, 24705, 
29052, 1891, 16454, 21256, 20375, 24932, 26868, 2047), resarea = c(14858870, 
14749327, 24253820, 23481299, 23925402, 2340073, 15931286, 17683557, 
23350742, 24030181, 26324026, 1817435), factryarea = c(4532, 
275223, 164443, 172154, 149287, 2006414, 4532, 54760, 301298, 
258694, 202702, 1052597), commarea = c(1681049, 4976927, 4074631, 
5551895, 3627362, 6017128, 1824355, 6283092, 5890039, 6682742, 
4507698, 5911856), officearea = c(129729, 204418, 462429, 1912298, 
231134, 530517, 129919, 215750, 692181, 1959893, 298734, 599141
), TotalLandValue = c(676413625, 889697539, 485997234, 904462580, 
1193787898, 326146766, 1216611479, 1568389461, 988172432, 1473477767, 
1486619757, 461667391), TotPop_p = c(50189, 61958, 48320, 69280, 
93751, 41346, 46009, 64686, 51867, 73806, 104715, 44024), TotPop_se = c(1544.96479999394, 
1932.91122991911, 1477.67713526889, 1621.11087850568, 2060.0971658131, 
1349.79788954655, 1463.73132791495, 1811.36511978071, 2126.19689193412, 
1625.67922346675, 2428.71826421752, 1460.56966131475), POC_p = c(0.66801490366415, 
0.768504470770522, 0.516804635761589, 0.337759815242494, 0.527898369084063, 
0.575630048855996, 0.701362776848008, 0.741968895897103, 0.586230165615902, 
0.37544373086199, 0.515962374062933, 0.534185898600763), POC_se = c(0.0214554297500083, 
0.0197231495256417, 0.0212115659870603, 0.0180918607533261, 0.0160111622396801, 
0.0254703302968768, 0.0249330451532724, 0.0205125632282244, 0.031076574811459, 
0.0161455000463029, 0.0143444833627841, 0.025013791232592), White_p = c(0.576898518414303, 
0.56703233256351, 0.372785800348877, 0.3746093375004, 0.419207632046567, 
0.276616843225463, 0.565697382933922, 0.562515242663198, 0.350836117682446, 
0.352671537029079, 0.382279196038988, 0.240164455751408), White_se = c(0.00675552182469777, 
0.00812057705141857, 0.00750732072656769, 0.0075838608821734, 
0.0070183352536629, 0.00975357052715687, 0.00744338106611418, 
0.00899123586507761, 0.00853192425238527, 0.00870241005563739, 
0.00554259946464601, 0.00817859134095142)), row.names = c(NA, 
-12L), class = c("tbl_df", "tbl", "data.frame"))

Suppose I wanted to calculate the % increment for the columns numbldgs, resarea and resunits, generating 3 new columns that can be differentiated from each other e.g. inc_bldgs, inc_resarea, inc_resunits.

Based on this question, it seems like mutate_at might be the way to go, but I am unsure about how to create the new fields with different names and how to set the generic function I need.

回答1:

Is it what you are looking for?

list(mod ) is used to add a suffix to the new columns created.

transform <- function(x) (100*(x-lag(x, default=first(x)))/lag(x, default=first(x)))


x %>% 
    arrange(CD) %>% 
    group_by(CD) %>% 
    mutate_at(c("numbldgs", "resarea", "resunits"),
              list(mod = transform))

# A tibble: 6 x 4
# Groups:   CD [3]
#  CD    numbldgs_mod resarea_mod resunits_mod
#  <chr>        <dbl>       <dbl>        <dbl>
# 1 103            0          0            0   
# 2 103           14.4        7.22         8.44
# 3 111            0          0            0   
# 4 111           13.3       19.9         15.1 
# 5 210            0          0            0   
# 6 210           17.3       -3.72         1.29

来源：https://stackoverflow.com/questions/61416762/mutate-df-to-generate-several-new-columns-using-a-function-on-several-original-f

标签

dataframe

dplyr

mutate