average calculus in C language

shifter · 06-30-2013, 03:51 PM

Hi all, I've a problem in average calculus with a program in C language.

Below I wrote some snippet of code.

In funzioni.c

Code:

float voto_medio(studente stud[], int dim) {

  int i, somma = 0;
  float vmedio;
  
  for (i = 0; i < dim; i++) {
    somma = stud[i].voto + somma;
  }

  vmedio = somma/dim;
  
  return (vmedio);
  
}

and in funzioni.h I have

Code:

#define MAX 50

typedef char STRINGA[30];

typedef struct {
  STRINGA nome;
  STRINGA cognome;
  int voto;
} studente;

void inserisci_studenti(studente*, int);
void visualizza_studenti(studente*, int);
float voto_medio(studente*, int);
void ordina(studente*, int);
void visualizza_studenti_con_giudizio(studente*, int);
void salva_su_file(studente*, int);

My problem is that average calculus in voto_medio function is not correct.
I'm not able to make average with float numbers. For example, with 30, 28 and 24 I have as result 27.000 whereas I must 27.333...

Where is my errore in voto_medio function?

Thanks in advance.

suicidaleggroll · 06-30-2013, 04:10 PM

somma and dim are ints, so somma/dim is performing an integer division. You'll need to re-cast them as floats in order to do a floating point division.

shifter · 06-30-2013, 04:22 PM

That is to say

Code:

(float)somma/dim;

modifying voto_medio in the following way:

Code:

float voto_medio(studente stud[], int dim) {

  int i, somma = 0;
  
  for (i = 0; i < dim; i++) {
    somma = stud[i].voto + somma;
  }
  
  return (float)somma/dim;
  
}

In this way I have average in floating point.

Thank you very much!!

H_TeXMeX_H · 07-01-2013, 03:18 AM

How accurate does the average need to be ? If it needs to be reasonably accurate, use 'double' instead of 'float'.

shifter · 07-02-2013, 11:04 AM

It doesn't matter, I think float is sufficient.

psionl0 · 07-02-2013, 07:16 PM

Quote:

Originally Posted by shifter

It doesn't matter, I think float is sufficient.

"float" is only guaranteed to be 32 bits long. About the only time you would use it in preference to "double" is when you are dealing with massive arrays of real numbers.

suicidaleggroll · 07-02-2013, 08:55 PM

Quote:

Originally Posted by psionl0

"float" is only guaranteed to be 32 bits long. About the only time you would use it in preference to "double" is when you are dealing with massive arrays of real numbers.

Or if you don't care about precision beyond 7 sig figs...

psionl0 · 07-03-2013, 12:14 AM

Quote:

Originally Posted by suicidaleggroll

Or if you don't care about precision beyond 7 sig figs...

You might not care whether you use float or double but you would seldom prefer float without a special reason.