Modeling human/computer interactions on Star Trek from #TidyTuesday with workflowsets | Julia Silge

[utterances-bot]

Modeling human/computer interactions on Star Trek from #TidyTuesday with workflowsets | Julia Silge

Learn how to evaluate multiple feature engineering and modeling approaches with workflowsets, predicting whether a person or the computer spoke a line on Star Trek.

https://juliasilge.com/blog/star-trek/

[SIRIYAK]

super cool video ,thanks insight😎🙏

[conlelevn]

@juliasilge about the sign (plus or minus) in estimate of SVM, does it have any particular meaning?

[juliasilge]

@conlelevn Yes, but it depends on which level of your outcome is "first". Here the "computer" level is first and the event of interest, while "people" is the second, non-event level. You can read more about this in the yardstick documentation for the relevant metrics.

[KleinWurm]

Hello, many thanks for that tutorial! May I ask a question?

What happens if we mix models with hyperparameter tuning and models without hyperparameters inside the workflowset? Especially at the workflow_map() step, the function fit_resample() might not be adequate. Do you have any tips on how to use this package in the right way in this case?

[juliasilge]

@KleinWurm If you use tune_grid() with a model that doesn't need tuning, it will just give you a message and then go ahead and fit to the resamples; that is what I would do in your case. Notice that is what happens in this chapter for bag_cart_spec.