使用dplyr对R中所有因变量进行分组汇总统计

Question

I am trying to generate groupwise (hearing - my independent variable, so HL and NH are the two groups) summary statistics (mean, sd, min, max, standard error etc. ) for each of the 10 dependent variables. I was able to do this for one variable (R_PTA) using these 2 codes:

1.

RightPTA <- mydata %>% group_by(NHL) %>% summarise(n=length(R_PTA), mean_R_PTA=mean(R_PTA), sd_R_PTA=sd(R_PTA), se_R_PTA=sd(R_PTA)/sqrt(length(R_PTA)), min_R_PTA=min(R_PTA), max_R_PTA=max(R_PTA))

2.

mydata
mean<-tapply(mydata$R_PTA, mydata$NHL, mean)
mean
sd<-tapply(mydata$R_PTA, mydata$NHL, sd)
sd
median<-tapply(mydata$R_PTA, mydata$NHL, median)
median
max<-tapply(mydata$R_PTA, mydata$NHL, max)
max
min<-tapply(mydata$R_PTA, mydata$NHL, min)
min
cbind(mean, sd, median, max, min)
round(cbind(mean, sd, median, max, min), digits = 1)
t1<-round(cbind(mean, sd, median, max, min), digits = 1)
t1

Here is the output:

RightearPTA
   mean  sd median  max min
HL 26.9 7.3   27.5 37.5 8.8
NH 11.6 4.1   12.5 16.2 2.5

I want the same exact thing for all the remaining 9 variables (L_PTA, B_PTA etc.) but in one shot if possible. Is there no way to do this? Do I have to code for each single dependent variable? I am sure its out there, but I cant find it! Any hep would be appreciated!!

Answer 1

Consider a base R solution with by (the object-oriented wrapper to tapply to subset dataframe into factor groups) and nested sapply (to build matrix of stats). Below demonstrates with random, seeded data for 10 stats columns:

set.seed(88)

df <- data.frame(
  GROUP = sapply(seq(50), function(i) sample(c("NH", "HL"), 1, replace=TRUE)),
  STAT1 = rnorm(50)*100,
  STAT2 = rnorm(50),
  STAT3 = runif(50)*100,
  STAT4 = runif(50),
  STAT5 = rgamma(50, shape = 2)*100,
  STAT6 = rgamma(50, shape = 2),
  STAT7 = rpois(50, lambda = 100)*100,
  STAT8 = rpois(50, lambda = 100),
  STAT9 = rexp(50, rate = 1)*100,
  STAT10 = rexp(50, rate = 1)
)

dfList <- by(df, df$GROUP, FUN = function(d)
                sapply(d[2:ncol(d)], function(i) 
                  c(mean = mean(i, na.rm=TRUE),
                    sd = sd(i, na.rm=TRUE),
                    median = median(i, na.rm=TRUE),
                    min = min(i, na.rm=TRUE),
                    max = max(i, na.rm=TRUE)
                  )
                )
            )

Output

dfList$HL

#              STAT1       STAT2     STAT3      STAT4     STAT5     STAT6      STAT7     STAT8      STAT9      STAT10
# mean     -6.594221 -0.04059519 52.990723 0.58753311 157.55220 1.9196911 10103.4483 101.17241 113.089148 0.771495372
# sd      102.512709  0.99159105 31.055376 0.27339871 152.37034 1.4880694   709.3673  10.02165 121.360898 0.720117072
# median    8.034055  0.01163562 56.416484 0.56894472 136.58274 1.5150241 10200.0000 103.00000  77.302150 0.599291434
# min    -199.786535 -1.84703449  1.345751 0.00207128  22.56936 0.1553518  8400.0000  82.00000   2.396641 0.006532798
# max     251.976970  2.55701655 98.612123 0.99413520 806.38484 7.1030277 11900.0000 120.00000 487.719745 3.133768953

dfList$NH

#             STAT1       STAT2      STAT3      STAT4    STAT5    STAT6      STAT7     STAT8      STAT9    STAT10
# mean     26.51853 -0.13748799 44.1973692 0.46621478 155.7555 1.880407  9961.9048 104.38095 150.596480 1.1243476
# sd       90.57645  0.77843518 29.9227560 0.30340507 121.5361 1.105004   868.6059   8.44083 131.123059 1.1627959
# median   24.52202 -0.02949522 46.1950960 0.33646282 114.7845 1.736198  9700.0000 105.00000 122.841835 0.7819896
# min    -105.54741 -1.58980314  0.2636007 0.02044767  17.3282 0.291350  8900.0000  89.00000   7.799051 0.1108107
# max     194.78958  1.35889041 96.0175463 0.99160167 434.5724 4.368176 12000.0000 120.00000 554.307036 5.1537741