getaero-io/gtm-eng-skills / skills/deepline-scoring
deepline-scoring
VerifiedUse when discovering niche signals, auditing ICP or won/lost evidence, rescoring accounts, or building account and lead scoring Plays. Triggers on fit scoring, engagement scoring, backtesting, holdout evaluation, predictive lift, external proxies, and scoring leakage. Skip pure outreach copy or contributor skill installation tasks.
scripts/scorecard/config.py
View on GitHubfrom __future__ import annotations
from pathlib import Path
from typing import List, Optional, Union, Literal
from pydantic import BaseModel, Field, model_validator
from typing import Dict
class GateRunConfig(BaseModel):
enable: bool = Field(default=False)
min_intrusion: float = Field(default=2.0)
min_prec_gain: float = Field(default=0.0)
class TiersRunConfig(BaseModel):
enable: bool = Field(default=True)
precA_min: float = Field(default=0.40)
covA_max: float = Field(default=0.12)
class PolicyRule(BaseModel):
id: str
target: str # column name (raw or engineered bin)
source: Literal["raw", "bin"] = Field(default="raw")
op: Literal["lte", "gte", "eq", "in", "between", "regex"] = Field(default="eq")
value: Optional[Union[float, str]] = Field(default=None)
values: Optional[List[Union[float, str]]] = Field(default=None)
override_points: int = Field(default=0)
enabled: bool = Field(default=True)
priority: int = Field(default=100)
@model_validator(mode="after")
def _validate_values(self) -> "PolicyRule":
if self.op == "between":
if not self.values or len(self.values) != 2:
raise ValueError("'between' op requires values: [low, high]")
if self.op == "in":
if not self.values or len(self.values) == 0:
raise ValueError("'in' op requires non-empty values list")
if self.op in ("lte","gte","eq","regex") and self.value is None:
raise ValueError(f"'{self.op}' op requires 'value'")
return self
class PolicyConfig(BaseModel):
rules: List[PolicyRule] = Field(default_factory=list)
combine: Literal["sum", "max"] = Field(default="sum")
class CatBinningConfig(BaseModel):
mode: Literal["te", "lift", "raw"] = Field(default="te")
columns: List[str] = Field(default_factory=list)
bucket_col_suffix: str = Field(default="_BIN")
drop_raw_cat: bool = Field(default=True)
lift_edges: Union[Literal["auto"], List[float]] = Field(default=[1.5, 3.0])
min_n: int = Field(default=50, ge=0)
min_pos: int = Field(default=3, ge=0)
zero_conv_min_n: int = Field(default=25, ge=0)
@model_validator(mode="after")
def validate_lift_edges(self) -> "CatBinningConfig":
if isinstance(self.lift_edges, list):
if len(self.lift_edges) != 2 or not (self.lift_edges[0] < self.lift_edges[1]):
raise ValueError("lift_edges must be 'auto' or two ascending floats, e.g. [1.5, 3.0]")
return self
class PreprocessConfig(BaseModel):
enable: bool = Field(default=False)
cat_binning: Optional[CatBinningConfig] = Field(default=None)
# numeric binning: create engineered categorical bins from numeric columns
numeric_bins: Dict[str, "NumericBinSpec"] = Field(default_factory=dict)
class RunConfig(BaseModel):
y_col: str
# "logistic" = propensity model (L1 logistic on WOE bins); "naive_bayes" = lift model (additive WOE points)
method: Literal["logistic", "naive_bayes"] = Field(default="naive_bayes")
woe_prior: float = Field(default=0.5, gt=0.0)
max_features: Optional[int] = Field(default=None, ge=1)
max_abs_points: Optional[int] = Field(default=None, ge=1)
test_size: float = Field(default=0.25, ge=0.0, le=0.9)
random_state: int = Field(default=42)
do_calibration: bool = Field(default=True)
n_bins_numeric: int = Field(default=5, ge=2)
n_bins_cat_high: int = Field(default=6, ge=2)
max_low_card: int = Field(default=25, ge=1)
l1_C: float = Field(default=1.0, gt=0.0)
scale: int = Field(default=10, ge=1)
top_share: float = Field(default=0.10, gt=0.0, le=1.0)
gate: GateRunConfig = Field(default_factory=GateRunConfig)
tiers: TiersRunConfig = Field(default_factory=TiersRunConfig)
policy: Optional[PolicyConfig] = Field(default=None)
# Optional: columns to pass through (identifiers) that should not be treated as model features
id_columns: Optional[List[str]] = Field(default=None)
class JobConfig(BaseModel):
csv: Path
out_dir: Path = Field(default=Path("tmp_reports/scorecard_out"))
include_cols: Optional[List[str]] = Field(default=None)
exclude_cols: Optional[List[str]] = Field(default=None)
run: RunConfig
preprocess: PreprocessConfig = Field(default_factory=PreprocessConfig)
@model_validator(mode="after")
def validate_columns(self) -> "JobConfig":
if self.include_cols and self.exclude_cols:
overlap = set(self.include_cols).intersection(set(self.exclude_cols))
if overlap:
raise ValueError(f"include_cols and exclude_cols overlap: {sorted(overlap)}")
return self
class NumericBinSpec(BaseModel):
edges: List[Union[float, str]]
unknown_bin: str = Field(default="Unknown")
bin_name: Optional[str] = Field(default=None)
drop_raw: bool = Field(default=True)
@model_validator(mode="after")
def _validate_edges(self) -> "NumericBinSpec":
if not self.edges or len(self.edges) < 2:
raise ValueError("numeric_bins.edges must provide at least two boundaries")
return self
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |