API Reference

Core Classes

BinningTransformer

from ScoreCardModel import BinningTransformer

Strategies: quantile, uniform, optimal, tree

Parameter	Type	Default	Description
`strategy`	str	`'quantile'`	Binning strategy
`n_bins`	int	`5`	Number of bins (ignored by `optimal`)
`variables`	list[str]	`None`	Subset of columns to bin
`bin_definitions`	dict[str, list[float]]	`{}`	Custom split points per feature

WOETransformer

from ScoreCardModel import WOETransformer

Methods: standard, adjusted, empirical_logit, signed, weighted

Parameter	Type	Default	Description
`method`	str	`'adjusted'`	WOE calculation method
`laplace_smoothing`	float	`1e-6`	Smoothing for `adjusted` method
`rare_lumping`	bool	`False`	Merge rare bins
`min_bin_pct`	float	`0.05`	Min bin population %
`rare_level_label`	str	`'RARE'`	Label for merged rare bins

Properties: - iv — Information Value per feature - woe_maps_ — WOE mapping per feature per bin

ScoreCardTransformer

from ScoreCardModel import ScoreCardTransformer

| Parameter | Type | Default | Description | |---|---|---|---|---| | model | Any | required | Fitted sklearn model | | binning_transformer | Any | required | Fitted BinningTransformer | | woe_transformer | Any | required | Fitted WOETransformer | | base_points | float | 600 | Base score | | base_odds | float | 50 | Odds at base score | | pdo | float | 20 | Points to double odds |

Methods: - transform(x) — Calculate scores - export_scorecard() — Export scorecard table - _repr_html_() — HTML table for Jupyter notebooks (no extra dependencies)

ScoreCardWrapper

from ScoreCardModel import ScoreCardWrapper

High-level facade combining binning, WOE, logistic regression, and scoring.

Parameter	Type	Default	Description
`binning_strategy`	str	`'quantile'`	Binning strategy
`n_bins`	int	`5`	Number of bins
`base_points`	float	`600`	Base score
`base_odds`	float	`50`	Odds at base score
`pdo`	float	`20`	Points to double odds

WOE Methods

Method	Use Case
`standard`	Baseline — `ln(%good / %bad)`
`adjusted`	Handles zero-count bins with Laplace smoothing
`empirical_logit`	Agresti correction for small datasets
`signed`	Preserves direction of rare events
`weighted`	Population-weighted WOE

Diagnostics

from ScoreCardModel.weight_of_evidence.diagnostics import (
    check_monotonicity, iv_by_bin, woe_chi_square,
    midpoint_correlation, bin_statistics
)

Analytics

Metrics

from ScoreCardModel.analytics.metrics import (
    calculate_ks, calculate_psi, calculate_accuracy_ratio
)

Plotting (16+ plot types)

from ScoreCardModel.analytics.plotting import (
    plot_ks, plot_roc, plot_cap, plot_gain_lift,
    plot_double_lift, plot_score_distribution, plot_calibration,
    plot_psi_drift, plot_woe_pattern, plot_iv_summary_enhanced,
    plot_event_rate_by_bin, plot_bin_stats, plot_iv_summary,
    plot_scorecard_waterfall, plot_scorecard_heatmap,
    plot_cutoff_optimization, plot_confusion_matrix
)

Feature Selection

from ScoreCardModel.analytics.selection import (
    select_by_iv, select_by_correlation, rank_features
)

Reporting

from ScoreCardModel.analytics.reporting import generate_report

Generates a Markdown report with embedded PNG plots, executive summary, and full scorecard table.

Interactive (Jupyter)

# Requires: pip install scorecard-toolkit[interactive]
from ScoreCardModel.interactive import ScorecardWidget
from ScoreCardModel.interactive import scorecard_to_html

Function / Class	Description
`ScorecardWidget`	Interactive what-if widget with sliders, live score, and waterfall chart
`scorecard_to_html(card_df)`	Convert scorecard DataFrame to styled HTML table

Templates

from ScoreCardModel.templates import BaseScorecard, ConservativeScorecard

Template	Description
`BaseScorecard`	Standard development with all features
`ConservativeScorecard`	Regulatory submissions, limited feature sets