Add missing metric config to sweep example in credit scorecards notebook

[johndmulhausen]

Bug

The sweep_config in colabs/boosting/Credit_Scorecards_with_XGBoost_and_W&B.ipynb defines method and parameters but no metric key. Without it, W&B sweep visualizations (parallel coordinates, parameter importance, best-run selection) pick an arbitrary parameter as the optimization goal instead of the validation loss the notebook actually cares about.

Fix

Insert a metric block into sweep_config, right after method:

  "metric" : {
    "name" : "val_log_loss",
    "goal" : "minimize"
  },

val_log_loss matches exactly what the sweep's train() function records (run.summary["val_log_loss"]), and it is a mean negative log-likelihood, so minimize is the correct goal. The block matches the cell's existing indentation and spacing style, and the change is a pure 4-line insertion in the raw notebook JSON — no other cells, outputs, or metadata touched.

Verification

• nbformat.validate() passes on the edited notebook (still valid nbformat 4 JSON).
• Extracted the edited cell's source (minus the wandb.sweep(...) call) and executed it: it is valid Python and sweep_config["metric"] == {"name": "val_log_loss", "goal": "minimize"}.
• Programmatically confirmed the sweep train() cell (the function passed to wandb.agent) still logs run.summary["val_log_loss"], so the metric name matches what training logs.

Fixes #601

🤖 Generated with Claude Code

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[Copilot]

Pull request overview

Adds the missing W&B sweep metric configuration to the Credit Scorecards hyperparameter sweep example so the sweep correctly optimizes validation log loss (instead of defaulting to an arbitrary parameter).

Changes:

• Inserted a "metric" block into sweep_config with name: "val_log_loss" and goal: "minimize".

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.