calculating an average, but removing lowest value

[jack_knuettel]

I have this dataset with 6 different hw grades. I ned to find the HW average, but I need the lowest value to be dropped. Here is the code:

 Name HW1 HW2 HW3 HW4 HW5 HW6

Abby	95.00	85.00	69.00	93.00	81.70	85.00
Hans	95.50	56.50	77.50	50.00	79.20	75.00
Beyonce	90.00	83.00	89.00	97.00	90.80	85.00
Camila	92.50	98.00	89.00	84.00	80.00	85.00
Carlos	95.00	69.00	55.00	92.00	76.70	100.00
Darius	95.00	97.00	99.00	89.00	100.00	87.50
Terrance	93.00	92.50	95.00	68.50	82.00	95.00
Evelyn	92.50	94.00	100.00	96.00	100.00	100.00
Gertrude	95.00	96.00	100.00	97.00	92.50	100.00
Alamance	93.50	84.00	100.00	91.00	72.50	85.00
Hiromi	85.50	94.00	85.00	84.00	97.50	100.00
Isla	80.00	40.00	45.00	61.70	64.00	72.50
Jasmine	95.00	98.00	100.00	96.00	100.00	100.00
Julieta	75.00	82.00	90.00	74.50	80.80	82.50
Lavanya	85.00	79.00	85.00	96.00	100.00	82.50
Oliver	98.00	100.00	96.50	98.00	97.50	100.00
Elon	83.00	98.00	87.50	96.00	100.00	75.00
Yvette	92.50	65.00	82.50	83.30	80.00	100.00

[ChrisNZ]

Lile this?

 data T;
  HW1=1;
  HW2=2;
  HW3=3;
  HW4=4;
  HW5=5;
  HW6=6;
  MIN=min(of HW1-HW6);
  AVG=mean(ifn(HW1=MIN,.,HW1)
          ,ifn(HW2=MIN,.,HW2)
          ,ifn(HW3=MIN,.,HW3)
          ,ifn(HW4=MIN,.,HW4)
          ,ifn(HW5=MIN,.,HW5)
          ,ifn(HW6=MIN,.,HW6));
run;

[PaigeMiller]

Method #1:

data want;
    set have;
    avg_without_min = (6*mean(of hw1-hw6) - min(of hw1-hw6))/5;
run;

Method B:

data want;
    set have;
    call sortn(of hw1-hw6);
    avg_without_min=mean(of hw2-hw6);
run;

To @jack_knuettel

What happens if there are two values of HW1-HW6 both with the minimum value (in other words, the two are tied)? Then what?

[ChrisNZ]

Or:

avg_without_min = (sum(of hw1-hw6) - min(of hw1-hw6))/(n(of hw1-hw6)-1);

[PaigeMiller]

---

@ChrisNZ wrote:

Or:

avg_without_min = (sum(of hw1-hw6) - min(of hw1-hw6))/(n(of hw1-hw6)-1);

---

Certainly an improvement on my earlier Method #1. Handles missing values properly as well.

[Reeza]

One more for fun, but won't scale at all.

array HW(6) HW1-HW6;

x = mean(largest(1, of HW(*)),
               largest(2, of HW(*)),
               ...
               largest(5, of HW(*));

Another method would be to transpose the data, sorting it then dropping the first record. 

I'd probably use @ChrisNZ solution in reality though.