Package index • scorecraft

Configuration

scr_config(): Pipeline configuration
scr_config_keys(): Dictionary of configuration keys
scr_presets(): Selection presets, side by side
scr_verbose(): Switch progress messages on or off

Stages

Every stage of the pipeline as a function, plus the shortcuts that chain them.

scr_align(): Stage 5: align a raw score to the declared scale
scr_bin(): Stage 2: optimal binning, screening, hold-out revalidation and pruning
scr_cutoff(): Stage 6: cut-off sweep with frozen cuts
scr_model(): Stages 3 and 4: multi-strategy selection and consensus
scr_reject(): Stage 6: honest reject inference through a sensitivity band
scr_scorecard(): Stages 4 and 5: points scorecard, aligned to the declared scale
scr_select(): Select variables for the scorecard
scr_split(): Stage 0: type the data and split train and hold-out
scr_strategy(): Stage 6: strategy table per band, with marginal expected profit
scr_triage(): Stage 1: descriptive triage and sentinel resolution

Coarse classing lab

Manual binning, manual variable choice and the decision ledger.

scr_classing_accept() scr_classing_discard(): Accept or discard a proposal
scr_classing_apply(): Commit the lab into a new selection result
scr_classing_choose(): Choose the final variable list manually
scr_classing_propose(): Propose manual bins for a variable
scr_classing_spec() scr_classing_read() scr_classing_import(): Classing specification as a long table, with its file round trip
scr_classing_view(): Inspect the current bins of a variable in the lab
scr_coarse_classing(): Coarse classing lab: manual binning and manual variable choice
scr_decisions(): Decision ledger of a lab, a result or a scorecard

Reading the result

scr_funnel(): Audit funnel: every input variable and its fate
scr_gains(): Gains table, at bin level
scr_leakage(): Leakage and suspicious-strength audit
print(<scr_result>) summary(<scr_result>) as.data.frame(<scr_result>) plot(<scr_result>): Result of a selection
scr_score_gains(): Score gains per frozen band
scr_score_metrics(): Score metrics per sample, with CI
scr_selected(): Variables approved for the scorecard

Production

Scoring in R, production SQL, deliverables.

predict(<scr_align>): Apply an alignment to raw scores
scr_apply(): Apply the WOE transformation or the scorecard to new data
scr_export(): Write the deliverables
scr_monitor(): Monitor the scorecard on new data
scr_monitoring_plan(): Monitoring plan read by scr_monitor()
scr_reasons(): Reason codes: the variables that took the most points from each row
scr_sql(): Production SQL

Metrics

scr_iv(): Information Value of any grouping
scr_metrics(): AUC, KS and Gini of a score, with a bootstrap confidence interval
scr_psi(): Population stability index, with the fixed and the sample-size-adjusted threshold

Score studies

Percentile bands, tiers, red / amber / green lights, claims about event rates, the operating point under constraints, two scores on the same rows, mix and rate effects, segments, maturity under censoring, uplift, rules against the score and time to detection.

scr_bands(): Percentile study of a score
scr_claims(): Probability statements about the event rate of score groups
scr_detection(): Time to detection of fraud episodes
scr_maturity(): Maturity of the event by score band
scr_mix_shift(): Mix and rate effects of a change in the event rate
scr_operating(): Operating point of a score under constraints
scr_overlap(): Overlap of rules and a score
scr_rag(): Red / amber / green lights of a score against its reference
scr_rag_plan(): Thresholds of the red / amber / green lights
scr_score_cross(): Two scores on the same rows
scr_segments(): One score on many segments
scr_tiers(): Tiers of a score: a few labeled levels of risk or propensity
scr_uplift(): Uplift of a treatment along a score

Portfolio and database

scr_compare(): Compare runs across targets
scr_core(): Variables that cross several targets
scr_run(): Run the selection for several targets straight from the database
print(<scr_runset>): Set of runs, one per target
scr_connect(): Connect to a database (ODBC DSN or any DBI driver)
scr_fetch(): Fetch a table with reproducible server-side sampling

IRB parameters and the default definition

Parameter tables by framework preset, the default engine and default rates by cohort.

scr_default(): Build the default flag from a monthly panel
scr_default_rate(): One-year default rates by cohort and the long-run average
scr_irb_params(): IRB parameter tables by framework preset

PD calibration and rating grades

predict(<scr_grades>): Grade a score vector with the cut points of an scr_grades object
predict(<scr_pd>): Predict grade and PD from an scr_pd object
scr_calibrate(): Calibrate the alignment to a central tendency
scr_grades(): Rating grades on the score
scr_master_scale(): Master scale of PD grades
scr_migration(): Migration matrix between two rating dates
scr_moc(): Margin of conservatism, by category
scr_pd(): The PD model: grades, margin of conservatism and the floor
scr_pd_pit_ttc(): One-factor bridge between point-in-time and through-the-cycle PD
scr_pd_validate(): Validate a PD model on a cohort panel

LGD

Workout loss given default, two-stage model, pools, downturn, floors and in-default estimates.

scr_elbe(): ELBE and in-default LGD on a grid of months since default
scr_lgd(): Two-stage LGD model and pools on the reference data set
scr_lgd_downturn(): Downturn LGD per pool
scr_lgd_floor(): Input floor on the downturn LGD per pool
scr_lgd_pools(): LGD pools from the predicted LGD
scr_lgd_validate(): Validation battery of an LGD model
scr_workout(): Workout LGD: the reference data set from default events and cash flows

EAD and credit conversion factors

scr_bin_continuous(): Bin drivers against a continuous target (LGD, CCF)
scr_ead(): Estimate CCF pools from the reference data set
scr_ead_data(): Build the realized-CCF reference data set from facility snapshots
scr_ead_downturn(): Downturn CCF per pool
scr_ead_validate(): Validate CCF pools: calibration, discrimination, back-testing and stability

Expected loss, capital and ECL

scr_capital(): Expected loss, risk-weighted assets and capital of a portfolio
scr_ecl(): Expected credit loss with stage allocation
scr_el(): Expected loss per exposure
scr_irb_rw(): IRB risk weight of one or many exposures
scr_pd_stress(): Stressed PD of the one-factor model
scr_sa_rw(): Standardized risk weight of an exposure

Data

scr_demo: Synthetic example data
scr_demo_ead: Synthetic monthly facility snapshots for the EAD/CCF module
scr_demo_lgd: Synthetic default events for the workout LGD examples
scr_demo_lgd_cashflows: Synthetic post-default cash flows of scr_demo_lgd
scr_demo_panel: Synthetic monthly panel for the default engine and PD calibration
scr_demo_portfolio: Synthetic exposure snapshot for expected loss, capital and ECL
scr_demo_rates: Synthetic monthly reference rate series