Should I normalize the signs of my input when computing the average?

Question

I'm trying to find the AVG of this array : 1742,1065,-67,-2988,-796,-1000,31,-67,-100,1180 I belive my method to change the number from POS to NEG is worng. my prog need's to calculate the above array then print massege if the AVG is NEG or POS, and it always print that the AVG is POS even though it's not. here is my code so far:

; lab56.asm
;
    .MODEL SMALL
    .STACK 100h
    .DATA
AVG_NEG  DB  'THE AVG IS NEG',13,10,'$'
AVG_POS  DB  'THE AVG IS POS',13,10,'$'
INDEX    DB  'Numbers that are larger than the average are in indexes:',13,10,'$'
RES      DB  '                     ','$'
ARR      DW  1742,1065,-67,-2988,-796,-1000,31,-67,-100,1180
Ten      DW  10
AVG      DW  0
temprint DB  '      ','$'
;Program start here:
    .CODE
    MOV AX,@DATA    ; DS can be written to only through a register
    MOV DS,AX       ; Set DS to point to data segment
    LEA SI, ARR
;
;   SUMUP
    MOV CX,10       ;10 variables in array
Sum:
    MOV AX,[SI]
    CMP AX,0
    JG Pos_label
    XOR AX,0000000000000000b
    ADD AX,0000000000000001b
Pos_label:
    ADD AVG,AX
    ADD SI,2        ;move to the next number
LOOP Sum
;   Divided by 10 to get the AVG
    CWD             ; AX -> DX:AX
    IDIV Ten
    MOV AVG,AX
;   print

;   Check if NEG or POS
    CMP AVG,0
    JG Avg_label
    MOV AH,9       ; Set print option for int 21h
    MOV DX,OFFSET AVG_NEG       ;  Set  DS:DX to point to AVG_NEG
    INT 21h
    JMP continue
Avg_label:
    MOV AH,9       ; Set print option for int 21h
    MOV DX,OFFSET AVG_POS       ;  Set  DS:DX to point to AVG_POS
    INT 21h
continue:
;

;Program end's here:

     MOV AH,4Ch       ; Set terminate option for int 21h
     INT 21h       ; Return to DOS (terminate program)
     END

Rob Kennedy Rob Kennedy · Accepted Answer · 2013-08-22T18:31:07

If you change the sign of all your inputs before you add them together, then you'll obviously always get that same sign for the sum (barring overflow), and thus for the average as well. Furthermore, the computed average will be wrong because you'll be averaging the magnitudes instead of the actual values.

Your method of changing the signs is indeed wrong, so don't do that. Just compute the average the same way you learned in grade school: Add the numbers and divide by the count.

Should I normalize the signs of my input when computing the average?

1 Answers