Robin Linacre

icon indicating a website link https://www.robinlinacre.com icon indicating an email [email protected]

icon company @moj-analytical-services icon location London icon location Data scientist/engineer for UK Government. Lead developer of Splink

Results 234 comments of Robin Linacre

(WIP) 2580 improve runtimes but pushing up common case statements into precomputed values

Is this a better implementation? Details ``` from __future__ import annotations from collections import Counter from typing import Dict, List, Set, Tuple from sqlglot import exp, parse_one # --------------------------------------------------------------------------- #...

(WIP) 2580 improve runtimes but pushing up common case statements into precomputed values

see https://github.com/moj-analytical-services/splink/pull/2738

Simple Splink Example

I think you're mixing up Splink 3 and Splink 4 syntax. The example on the page you link is: ``` import splink.comparison_library as cl from splink import DuckDBAPI, Linker, SettingsCreator,...

Simple Splink Example

All the docs relate to the latest version - Splink 4. The banner at the top should probably be removed now, it's just for users maintaining old Splink 3 code

Errors running development_setup_with_conda.sh and make_docs_locally.sh

Yeah, that's right. Upgrading the docs to Splink 4 was a huge tasks and I lost the motivation to do everything. Also there's better error messages now, so the tables...

[Athena] calling `invalidate_cache()` results in "not a folder created by Splink" error even when folder was created by Splink

Sorry you're having problems. We're probably not going to have time to look at this any time soon. If you can find the bug, feel free to PR a fix...

[FEAT] add option to generate match weights chart off linker

I've been mulling this over. I had forgotten about this, but you can actually achieve this functionality as follows: ``` SettingsCreator.from_path_or_dict(path).get_settings('duckdb').match_weights_chart() ``` This is a bit of a hangover from...

[FEAT] Add test coverage

This could also be useful to identify any code that is no longer actually used in Splink

Possible bug in `estimate_u_using_random_sampling` for Spark backend

Thanks for the report. Yeah, the solution of materialising the sampled table sounds sensible to me and should have very little impact on performance

Evaluation from ground truth column does not work without blocking rules specified

Thanks for the report - yeah, this def looks like something we should fix