Skip to content
RiskBands RiskBands RiskBands

Examples

Start here

Section titled “Start here”

API ergonomics notebook

Section titled “API ergonomics notebook”

This is the recommended entry point for learning RiskBands with a workflow that feels closer to pandas and sklearn.

This material shows:

  • fit(df, y="target", column="score", time_col="month")
  • transform(df["score"])
  • summary()
  • binning_table()
  • score_details()
  • diagnostics()
  • comparison between standard and stable

Temporal stability quickstart

Section titled “Temporal stability quickstart”

This flow shows score_table(), audit_table(), JSON/bundle export, and public plots for temporal reading.

Missing policy with pandas and PySpark

Section titled “Missing policy with pandas and PySpark”

Use these scripts when the main question is how to handle missing values in an auditable way without opaque imputation.

They show:

  • missing_policy="standard"
  • missing_policy="separate_bin"
  • missing_policy="forbid"
  • missing_policy="merge" with nearest_event_rate
  • missing_policy="merge" with nearest_woe
  • missing_profile_
  • missing_decision_log_
  • missing_merge_candidates_
  • bundle fields for missing policy and missing merge
  • optional guard for PySpark
  • comparison across standard, separate_bin, forbid, merge + nearest_event_rate, and merge + nearest_woe
  • synthetic credit-risk data without real or sensitive records

Narrative audit report

Section titled “Narrative audit report”

This flow shows:

  • a small synthetic dataset;
  • missing_policy="merge";
  • missing_merge_criterion="nearest_event_rate";
  • export_audit_report("audit_report.html");
  • export_bundle("bundle") with audit_report.html included by default;
  • generated file paths without opening a browser automatically.

PD vintage champion/challenger

Section titled “PD vintage champion/challenger”

PD vintage benchmark

Section titled “PD vintage benchmark”

Use this material when the main question is why a candidate with stronger aggregate IV can still be the wrong choice for credit when time enters the decision.

stable score demo

Section titled “stable score demo”

Suggested reading order

Section titled “Suggested reading order”

If you want to start with the API

Section titled “If you want to start with the API”
  1. Synthetic notebook with Plotly
  2. Quickstart
  3. API overview
  4. Outputs and diagnostics
  5. Missing policy
  6. Narrative audit report
  7. PD vintage champion/challenger
  8. PD vintage benchmark

If you want to start with the methodology

Section titled “If you want to start with the methodology”
  1. Why RiskBands
  2. Why not only OptimalBinning
  3. PD vintage benchmark
  4. How to read the charts