[scikit-learn] methods available from last estimator in pipeline

[Gregory, Matthew]

Ben and Sole,

Thanks for your helpful suggestions and insight.  Ben, I appreciate the additional details on what methods carry over and the nice shorthand on how to apply in case they don’t carry over.

matt

From: bmreiniger <bmreiniger at yahoo.com>
Sent: Saturday, September 24, 2022 7:11 AM
To: Sole Galli via scikit-learn <scikit-learn at python.org>
Cc: Sole Galli <solegalli at protonmail.com>; Gregory, Matthew <matt.gregory at oregonstate.edu>
Subject: Re: [scikit-learn] methods available from last estimator in pipeline


[This email originated from outside of OSU. Use caution with links and attachments.]
Matt would still need to transform the data using the rest of the pipeline first. There are multiple naming options for pipeline steps, but in this case indexing is nice:

```
pipe[-1].kneighbors(
    pipe[:-1].transform(X)
)
```

The user guide quote should probably be updated: not every method carries over. In the source (https://github.com/scikit-learn/scikit-learn/blob/36958fb240fbe435673a9e3c52e769f01f36bec0/sklearn/pipeline.py#L426) you can see what does, indicated by the `@available_if(_final_estimator_has(...))` decorator. These are the same as those listed in the docs page (https://scikit-learn.org/stable/modules/generated/sklearn.pipeline.Pipeline.html#sklearn.pipeline.Pipeline.decision_function). Anything else you need to call from the last step itself.

Best,
Ben Reiniger


-------- Original message --------
From: Sole Galli via scikit-learn <scikit-learn at python.org<mailto:scikit-learn at python.org>>
Date: 9/24/22 5:56 AM (GMT-06:00)
To: Scikit-learn mailing list <scikit-learn at python.org<mailto:scikit-learn at python.org>>
Cc: Sole Galli <solegalli at protonmail.com<mailto:solegalli at protonmail.com>>
Subject: Re: [scikit-learn] methods available from last estimator in pipeline

Did you try:

pipeline.named_steps["the_string_name_for_knn"].kneighbours

?

pipeline should be replaced by the name you gave to your pipeline and the string in named_steps is the name you have to the knn when setting the pipe.

Sole

Sent with Proton Mail secure email.

------- Original Message -------
On Friday, September 23rd, 2022 at 10:16 PM, Gregory, Matthew <matt.gregory at oregonstate.edu<mailto:matt.gregory at oregonstate.edu>> wrote:


> Hi all,
>
> I have what is probably a silly question. I read this passage on [1]:
>
> """
> The pipeline has all the methods that the last estimator in the pipeline has, i.e. if the last estimator is a classifier, the Pipeline can be used as a classifier. If the last estimator is a transformer, again, so is the pipeline.
> """
>
> I'm trying to create a pipeline where my last estimator is a KNeighborsClassifier and, instead of predict(), I was hoping to use kneighbors(). But unfortunately, when in a pipeline, I'm getting this AttributeError:
>
> AttributeError: 'Pipeline' object has no attribute 'kneighbors'
>
> Is kneighbors() really available from the Pipeline? Or is there an alternative way to call an element in the Pipeline to use it? I tried "pipe[-1].kneighbors(X)", but that doesn't seem to be applying the earlier transforms in the pipeline.
>
> Thanks for any pointers,
> matt
>
> [1] https://scikit-learn.org/stable/modules/compose.html
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org<mailto:scikit-learn at python.org>
> https://mail.python.org/mailman/listinfo/scikit-learn
_______________________________________________
scikit-learn mailing list
scikit-learn at python.org<mailto:scikit-learn at python.org>
https://mail.python.org/mailman/listinfo/scikit-learn
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20220926/4321a182/attachment.html>