In D-Se/ML: Machine Leanring targets-bookdown pipeline

(PART) Model performance {.unnumbered}

Metrics

• Introduce metrics here

• ROC

• AUC
• RMSE
• MAE
• Precision
• Accuracy

Model performance

Here is an example constructed from sample data, training 12 different ML models.

::: question

Which model performed the best? :::

library(ggplot2)
targets::tar_read(ranking) +
  theme_bw()

A quick overview, obtained from a node in the pipeline.

::: note

Always check the actual predictions to look for anomalies! :::

And to check if the predictions are close to the actual observations, we inspect another node from the pipeline.

targets::tar_read(visual_verify) +
  theme_bw()

Looks good. What is the error metric associated with this?

targets::tar_read(metrics)

::: question

Looks neat! how do I get this model in my RStudio to test it? What are the model parameters? :::

That is the benefit of making your own package.
- Simply run ML::results to get the best performing model straight into your R session - Look at the documentation by running ?results.

ML::results

D-Se/ML documentation built on April 1, 2022, 10:53 p.m.