Closed Cyeee closed 4 years ago
I modified the code a bit and it seems this will work.
` alphabets_in_sequence = np.unique(sequence)
for i, u in enumerate(alphabets_in_sequence):
index = [p[0] for p in positions].index(u)
U = np.array(positions[index][1]).ravel()
for j, v in enumerate(alphabets_in_sequence):
index = [p[0] for p in positions].index(v)
V2 = np.array(positions[index][1]).ravel()
C = [(i,j) for i in U for j in V2 if j > i]
## yi added
pos_i = alphabets_list.index(u)
pos_j = alphabets_list.index(v)
W0[pos_i,pos_j] = len(C)
## yi added end
cu = np.array([i[0] for i in C])
cv = np.array([i[1] for i in C])
## yi added
Wk[pos_i,pos_j] = np.sum(np.exp(-kappa * np.abs(cu - cv)))
## yi added end`
alphabets_list = list(alphabets)
I agree this seems to be incorrect. Basically the indexes of W0 and Wk can change if the alphabet is not entirely in the sequence.
alphabets_in_sequence = np.unique(sequence)
for i, u in enumerate(alphabets_in_sequence):
@Cyeee @gabrielelanaro I am reviewing this. You can expect a resolution in a week. In the meantime, there is an R package on my github repo that you can use.
@Cyeee @gabrielelanaro The issue is resolved. Please
pip uninstall sgt
. pip install sgt==1.0.0
.Thanks for fixing this!
Thanks for fixing it! Yi
Sent from my iPhone
On Dec 17, 2019, at 4:34 PM, cran2367 notifications@github.com wrote:
@Cyeee @gabrielelanaro The issue is resolved. Please
uninstall your current sgt using pip uninstall sgt. Reinstall with pip install sgt==1.0.0. Run the notebook here: https://github.com/cran2367/sgt/blob/master/python/sgt.ipynb If any problem is still there please comment. If the problem is resolved, please close the issue. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.
Hi Chitta Ranjan, Thanks so much for developing this method/package. It is super helpful to me and is driving real business impact on my work. One issue I noticed is that in the python package, it gave same results for sequence ['a','b'] and ['e','f']. I assume the results should be different? Many thanks again for developing this! Yi