Paranthesis mistake with assortativity_bin and assortativity_wei about bctpy HOT 3 CLOSED

Hi,

I have no doubt your formulation is clearer than the code currently (which
is directly modeled off of BCT matlab code) but can you point out
explicitly where is the error in parenthesis?

On Tue, Mar 31, 2015 at 5:18 PM, birkantunc [email protected]
wrote:

Hi,

For both assortativity_bin and assortativity_wei, there is a mistake in
calculation of "r"

Right formulation is

term1 = np.sum(degi
degj)/K term2 = np.square(np.sum(.5(degi+degj))/K)
term3 = np.sum(.5_(degi_degi+degj*degj))/K
r = (term1 - term2) / (term3 - term2)

for binary and

term1 = np.sum(stri
strj)/K term2 = np.square(np.sum(.5(stri+strj))/K)
term3 = np.sum(.5_(stri_stri+strj*strj))/K
r = (term1 - term2) / (term3 - term2)

for weighted matrices.

Regards,
Birkan

—
Reply to this email directly or view it on GitHub
#7.

from bctpy.

Hi,

Sorry for not giving the details.

In the original form it is

r=(( np.sum(degi_degj)/K - np.square(np.sum(.5_(degi+degj))/K)) /
np.sum(.5_(degi_degi+degj_degj))/K - np.square(sum(.5_(degi+degj))/K))

It should be

r=( np.sum(degi_degj)/K - np.square(np.sum(.5_(degi+degj))/K)) /
(np.sum(.5_(degi_degi+degj_degj))/K - np.square(sum(.5_(degi+degj))/K))

i.e. first parenthesis should be moved to the beginning of the denominator. The correct form is
(t1 - t2) /(t3 - t4)
but currently it is
((t1 - t2) / t3 - t4)
Thus, results are not in the range of [-1, 1] as it should be. Again, since always t4 = t2, you do not need to calculate it twice.

from bctpy.

You are right, thanks.

from bctpy.