Tuning hyperparameters and stacking models with 'tidymodels' | R Tutorial (2021)

Caret tutorial series:

Tidymodels:

In the previous tutorial series, we walked through the "caret" package in R for machine learning. We used the raw "GermanCredit" dataset, performed a brief exploration of it, and used the package to walk through a variety of steps: pre-processing, removing low information features, tuning hyperparameters, correcting for class imbalance, and summarizing results based on metrics we deem important. Where possible, we will perform the exact same exercise here, except we will use the "tidymodels" suite of packages to do so.

There are a few sources from which this tutorial draws influence and structure.

[hungcheung]

After watching the tutorials on both, I find caret far easier to understand/explain and simpler to set up. tidymodels doesn't work for me. At best it's only better for the recipes part. Just my thoughts, I find the pipe is good for incremental actions which you can read through whets going on the entirety of the length of the pipe (start>finish) unbroken. Here the pipe draws in complex settings setup outside the pipe. I'm doing a lot of mental gymnastics to keep track. Btw, thanks for a great series of videos, very comprehensive

[dijanaostojic]

Hi Richard, thanks so much for your videos; they've been incredibly helpful! I’m currently working with a highly imbalanced dataset—1% positive class and 99% negative class—and I'm interested in adjusting the classification threshold using tidymodels. I’ve read that direct threshold modification might not be supported yet. Is there any workaround for this, or any alternative methods you recommend for handling this kind of class imbalance? I’d appreciate any advice or resources you could share!

[AbdullahN]

I am new to tidy models. Is it worth to try to learn Caret also?
and thanks a lot for this amazing channel

[squaretube]

Thanks Richard for the amazing videos 👏👏

[fullsurr]

Richard, many thanks for the tidyverse series! I watched all three with pleasure + english idioms which are new to me like "meat and potato"). Seriously, three rows of code in caret vs "pages" you've just showed, couple of times you have even wrinkled up and I do think I understand why... Does the ensembling part in caret so bad and does it worth using tidy just because of it? Am i missing anything? Where does tidy beat caret and what do you use more frequently in real-life problems?

[ekvillain]

1. I want to become a data scientist what do i need to learn?
2.for data scientist sql is better or mysql? Plz help me 🙃