Get an error message while running match_string about string_grouper

[ParticularMiner]

Hi @iibarant

please provide more information. 88510 is small enough to handle by my computer (which can process >500,000 records).

[iibarant]

Here's the code:

from string_grouper import match_strings
matches = match_strings(check2['full address'])

I run the code on MacBook Pro 2.4 GHz 8-Core Intel Core i9 32 GB 2667 MHz DDR4
The dataframe contains 3 columns name, phone, full address and I need to run the match on the address only.

Thank you!

[ParticularMiner]

Thanks @iibarant

Curious! This is an unexpected error. Can you please provide the traceback log (just copy and paste whatever python spits out) of the error so that I can determine where exactly the problem is stemming from in the code.

[iibarant]

There you go ...

matches = match_strings(check2['full address'])
Traceback (most recent call last):

File "", line 1, in
matches = match_strings(check2['full address'])

File "/opt/anaconda3/lib/python3.8/site-packages/string_grouper/string_grouper.py", line 131, in match_strings
string_grouper = StringGrouper(master,

File "/opt/anaconda3/lib/python3.8/site-packages/string_grouper/string_grouper.py", line 264, in fit
matches, self._true_max_n_matches = self._build_matches(master_matrix, duplicate_matrix)

File "/opt/anaconda3/lib/python3.8/site-packages/string_grouper/string_grouper.py", line 467, in _build_matches
return awesome_cossim_topn(

File "/opt/anaconda3/lib/python3.8/site-packages/sparse_dot_topn/awesome_cossim_topn.py", line 119, in awesome_cossim_topn
alt_indices, alt_data = ct_thread.sparse_dot_topn_extd_threaded(

File "sparse_dot_topn/sparse_dot_topn_threaded.pyx", line 133, in sparse_dot_topn.sparse_dot_topn_threaded.__pyx_fuse_0sparse_dot_topn_extd_threaded

File "sparse_dot_topn/sparse_dot_topn_threaded.pyx", line 168, in sparse_dot_topn.sparse_dot_topn_threaded.sparse_dot_topn_extd_threaded

OverflowError: value too large to convert to int

[ParticularMiner]

@iibarant

Looks like the error stems from ‘sparse_dot_topn’, a package dependency.
And I’m not sure if it’s platform-dependent (the package, as far as I know, has only been tested on Linux and Microsoft Windows OS’s) or something else.

Could you try the following command just to see what happens:

matches = match_strings(check2['full address'], max_n_matches=20)

(This limits the output a bit.)

[iibarant]

Yes, that works. Thank you. Should I check whether the code works with greater max_n_matches ? I'm planning to keep similarity score > 0.9. Would it be possible to apply on the call?

[ParticularMiner]

@iibarant

Ok good. Yes, I suggest you try successively larger values of max_n_matches until the output size no longer changes. Yes of course, you can also use min_similarity at the same time.