Feedback on tutorial

[staropram]

Hi, first of all let me say thanks for creating such an amazing package with comprehensive documentation and tutorials. You really have all done an excellent job on it. Go Team. I've gone through the tutorial 3 times and as such have found a couple of typos and have some suggestions.

1. For the getting started page (or perhaps as intro to tutorial), to encourage good library hygene and to ensure things work first time for new users, could suggest a workflow like this:

python3 -m venv ~/venvs/splink_tutorial
. ~/venvs/splink_tutorial/bin/activate
pip install splink notebook
jupyter notebook

Then they'd have a notebook setup that should just work. Perhaps that could be at the start of the tutorial and this might reduce support requests coming in with broken setups.

2. in the "Blocking" part in the section entitled "Counting the number of comparisons created by a single blocking rule" it references the SQL of the blocking rules examined.

blocking_rule_1.blocking_rule_sql

this should be

blocking_rule_1.sql

etc

3. In the "Estimating model parameters" section
   some of the the code examples have retain escaped sequences for < such as &lt; so the code examples won't copy and paste without modification

same issue in "Clustering" in the "Predicting results" section

4. In "Quality assurance" section, the "load in labels" bit should probably put df_labels.head(5) as the last line so it is shown in the notebook

That's it. Amazing stuff, cheers!

[RossKen]

Hey @staropram - thanks so much for the feedback! I will open up a PR now to make these changes 😊

[staropram]

Cheers!

[RobinL]

@staropram just to note

blocking_rule_1.blocking_rule_sql this should be blocking_rule_1.sql

That's my fault. We have recently made a change so that in future it will be blocking_rule_1.blocking_rule_sql but in the version of code you're using it's not yet. (If you use Splink on master, or wait for the next release, that's what it will be)

Only noting to say sorry for the trouble, and if you see Ross hasn't fixed this, this is why (he's not ignoring the feedback)

[staropram]

Ah I see, makes sense, thanks for the clarification. Would you recommend using Splink on master?

[RobinL]

There's no problem in doing so, but I' not aware of any major bugs or features that have recently been released, so I don't think you'll see any big practical differences in terms of functionality

The change in blocking_rule_sql is a 'under the hood' thing to make it easier for Splink developers to understand the backend code

[staropram]

Thanks