Quantcast
Channel: Active questions tagged r - Stack Overflow
Viewing all articles
Browse latest Browse all 201867

How to obtain percentile of variable and the rate of dummy for all

$
0
0

How to do this for all columns of the object x automatically, using base R or SAS base.

Here is example using R:

# sample data

set.seed(123)
x <- data.frame(var1=runif(100), var2=runif(100), flag=rbinom(100, size=1, prob=0.7))
x

# calculate percentile of each column 

r <- apply(x, 2, function(x) quantile(x, probs=seq(0,1,0.05)))
res <- data.frame(item_id=rownames(r), r, row.names = NULL)

# assign group for each percentile

res$group <- seq_along(res$item_id)
res

# find the bin of the variable (var1, var2, ...) given percentile bin (interval);
x$bin_var1 <- findInterval(x$var1, res$var1)
x

# calculate the occurence, rate of the dummy flag column name (no=no occurence; yes=occurence of flag==1; total=total obs per bucket; rate_var=rate of var1)
op <- data.frame(with(x, aggregate(flag, list(bin_var1), FUN=function(x) c(sum(x==0),sum(x==1), length(x), sum(x==1)/length(x)))))
op1 <- data.frame(do.call(data.frame, op))
colnames(op1) <- c("group","no","yes","total","rate_var1")
op1

# merge
final <- merge(res, op1, by="group")
final

In this SAS solution I'm missing how to include the rate the ration of flag=1/flag all, in R I'm using the findInterval function to assign bin and then calculate the rate, sum(flag=1)...this part I'm not sue how to do in SAS.

Example:

data x;
length groups $12;
input groups Var1 Var2 Flag;
datalines;
constrict       3.50  1.09   1 
constrict       0.75  1.50   0
constrict       0.70  3.50   1
no_constrict    1.10  1.70   1
no_constrict    0.90  0.45   1
no_constrict    0.55  2.75   1 
no_constrict    1.40  2.33   0
constrict       2.30  1.64   1
constrict       0.85  1.415  0
no_constrict    1.80  1.80   1
no_constrict    0.95  1.36   1
no_constrict    1.50  1.36   0
constrict       0.60  1.50   0 
constrict       0.95  1.90   0
constrict       1.60  0.40   1
constrict       2.35  0.03   1
no_constrict    1.10  2.20   0
constrict       0.80  3.33   0
no_constrict    0.75  1.90   0
;


proc univariate data=x noprint;
class groups;
var var1
    var2
    flag
    ;
output out=res pctlpts=0 to 100 by 10 pctlpre=var1_ 
    var2_
    flag_
   ;


proc sql;
  create table op1
  as select a.*, 
             b.*
  from x 
  as a 
  left join res
  as b 
  on a.groups=b.groups;
quit;

Viewing all articles
Browse latest Browse all 201867

Trending Articles



<script src="https://jsc.adskeeper.com/r/s/rssing.com.1596347.js" async> </script>