code to find correlation

G

Gary Wessle

Hi

I wrote a code to calculate the correlation between 2 sequence of
double numbers presented as vectors.
did I go about this the right way?

thanks

#include <vector>
using std::vector;
#include <valarray>
using std::valarray;


cor::cor(vector<double>& x, vector<double>& y)
{
int c = x.size();
int d = y.size();
if(c != d)
cout << "to calculate correlation, variables"
" must be of the same number of elements" << endl;

// converts the input to valarray for spead and ellegance.
valarray<double> a(c);
valarray<double> b(d);
copy(x.begin(), x.end(), &a[0]);
copy(y.begin(), y.end(), &b[0]);

// expected values "mean"
double Ea = a.sum() / static_cast<double> (a.size());
double Eb = b.sum() / static_cast<double> (b.size());
// standard diviation
double Sa = sqrt( pow((a-Ea),2).sum()/c );
double Sb = sqrt( pow((b-Eb),2).sum()/d );
// double Sb = sqrt( ((a*a).sum()/c)- pow(Ea, 2) ); // faster
// but more round-offs

// correlation
cof = ( ((a-Ea)/Sa)*((b-Eb)/Sb) ).sum()/(c-1);
}
 
V

Victor Bazarov

Gary said:
I wrote a code to calculate the correlation between 2 sequence of
double numbers presented as vectors.
did I go about this the right way?

I would never use 'cout' in a constructor unless it's a debug output
(in that case it should go to 'cerr' and be #ifdef'ed). You might
think of throwing an exception instead.
thanks

#include <vector>
using std::vector;
#include <valarray>
using std::valarray;


cor::cor(vector<double>& x, vector<double>& y)
{
int c = x.size();
int d = y.size();
if(c != d)
cout << "to calculate correlation, variables"
" must be of the same number of elements" << endl;

// converts the input to valarray for spead and ellegance.
valarray<double> a(c);
valarray<double> b(d);
copy(x.begin(), x.end(), &a[0]);
copy(y.begin(), y.end(), &b[0]);

// expected values "mean"
double Ea = a.sum() / static_cast<double> (a.size());
double Eb = b.sum() / static_cast<double> (b.size());
// standard diviation

Isn't it "deviation"?
double Sa = sqrt( pow((a-Ea),2).sum()/c );
double Sb = sqrt( pow((b-Eb),2).sum()/d );

What are the extra parentheses aroud 'x-Ex' for?
// double Sb = sqrt( ((a*a).sum()/c)- pow(Ea, 2) ); // faster
// but more round-offs

// correlation
cof = ( ((a-Ea)/Sa)*((b-Eb)/Sb) ).sum()/(c-1);
}

Well, I see 'cor' is an object. Why? Does it have states? Does
it have a life on its own? What's the reason for it to be an object?
Shouldn't you simply have a function, like

double cor(vector..., vector...)
{
...
return cof;
}

?

V
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Members online

No members online now.

Forum statistics

Threads
473,967
Messages
2,570,148
Members
46,694
Latest member
LetaCadwal

Latest Threads

Top