File size: 7,212 Bytes
5db8ecb | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 | from __future__ import annotations
import math
from datetime import date, datetime, timedelta
from pathlib import Path
from typing import Any, Mapping
def resolve_schema_path(base_dir: Path) -> Path:
candidates = [
base_dir / "model" / "schema" / "runtime_schema_DE.json",
base_dir / "model" / "resources-to-build" / "runtime_schema_DE.json",
]
for candidate in candidates:
if candidate.exists():
return candidate
searched = ", ".join(str(path.relative_to(base_dir)) for path in candidates)
raise FileNotFoundError(f"Runtime schema not found. Checked: {searched}")
def to_float_or_none(value: Any) -> float | None:
if value is None or value == "":
return None
try:
number = float(value)
return number if math.isfinite(number) else None
except (TypeError, ValueError):
return None
def to_categorical(value: Any) -> str:
if value is None:
return "MISSING"
text = str(value).strip()
return text if text else "MISSING"
def parse_date(value: Any) -> date | None:
if value is None:
return None
if isinstance(value, date):
return value
for fmt in ("%Y-%m-%d", "%Y/%m/%d", "%d-%m-%Y", "%d/%m/%Y"):
try:
return datetime.strptime(str(value).strip(), fmt).date()
except ValueError:
continue
return None
def periodo_to_fecha_corte(periodo: int | str) -> date | None:
value = str(periodo).strip()
if len(value) != 6 or not value.isdigit():
return None
year, month = int(value[:4]), int(value[4:])
if not 1 <= month <= 12:
return None
if month == 12:
return date(year, 12, 31)
return date(year, month + 1, 1) - timedelta(days=1)
def years_diff_floor(start: date | None, end: date | None) -> int | None:
if start is None or end is None:
return None
years = end.year - start.year
if (end.month, end.day) < (start.month, start.day):
years -= 1
return years
def months_diff_floor(start: date | None, end: date | None) -> int | None:
if start is None or end is None:
return None
months = (end.year - start.year) * 12 + (end.month - start.month)
if end.day < start.day:
months -= 1
return months
def safe_log1p(value: Any) -> float | None:
number = to_float_or_none(value)
if number is None:
return None
return math.log1p(max(number, 0.0))
def cap_dias_mora(dias_mora: Any) -> float | None:
value = to_float_or_none(dias_mora)
if value is None:
return None
return min(value, 120.0)
def bucket_mora(dias_mora: Any) -> str:
value = to_float_or_none(dias_mora)
if value is None:
return "MISSING"
if value <= 0:
return "0"
if value <= 8:
return "1_8"
if value <= 30:
return "9_30"
if value <= 60:
return "31_60"
if value <= 89:
return "61_89"
return "90_plus"
def horizonte_hasta_dic(fecha_snapshot: date | None) -> int | None:
if fecha_snapshot is None:
return None
return 12 - fecha_snapshot.month
def safe_ratio(numerator: Any, denominator: Any) -> float | None:
left = to_float_or_none(numerator)
right = to_float_or_none(denominator)
if left is None or right is None or right <= 0:
return None
return left / right
def build_feature_map(raw: Mapping[str, Any]) -> tuple[dict[str, Any], dict[str, Any]]:
fecha_snapshot = parse_date(raw.get("fecha_snapshot")) or periodo_to_fecha_corte(raw["periodo"])
fecha_nacimiento = parse_date(raw.get("fecha_nacimiento"))
fecha_1er_desembolso = parse_date(raw.get("fecha_1er_desembolso"))
fecha_vencimiento = parse_date(raw.get("fecha_vencimiento"))
saldo_actual = to_float_or_none(raw.get("saldo_actual"))
capital_desembolsado = to_float_or_none(raw.get("capital_desembolsado"))
cuota_capital = to_float_or_none(raw.get("cuota_capital"))
dias_mora = to_float_or_none(raw.get("dias_mora_capital"))
dias_mora_capped = cap_dias_mora(dias_mora)
derived = {
"fecha_snapshot": str(fecha_snapshot) if fecha_snapshot else None,
"horizonte_meses_hasta_dic": horizonte_hasta_dic(fecha_snapshot),
"edad_cliente_t": years_diff_floor(fecha_nacimiento, fecha_snapshot),
"edad_credito_meses_t": months_diff_floor(fecha_1er_desembolso, fecha_snapshot),
"plazo_original_meses": months_diff_floor(fecha_1er_desembolso, fecha_vencimiento),
"plazo_remanente_meses_t": months_diff_floor(fecha_snapshot, fecha_vencimiento),
"dias_mora_capital_capped": dias_mora_capped,
"mora_bucket": bucket_mora(dias_mora),
"ratio_saldo_capital": safe_ratio(saldo_actual, capital_desembolsado),
"ratio_cuota_saldo": safe_ratio(cuota_capital, saldo_actual),
"ratio_cuota_capital_des": safe_ratio(cuota_capital, capital_desembolsado),
"saldo_actual_log": safe_log1p(saldo_actual),
"capital_desembolsado_log": safe_log1p(capital_desembolsado),
"cuota_capital_log": safe_log1p(cuota_capital),
}
feature_map = {
"id_cooperativa": to_categorical(raw.get("id_cooperativa")),
"calificacion_credito": to_categorical(raw.get("calificacion_credito")),
"saldo_actual": saldo_actual,
"capital_desembolsado": capital_desembolsado,
"cuota_capital": cuota_capital,
"tasa_interes": to_float_or_none(raw.get("tasa_interes")),
"tasa_pactada": to_float_or_none(raw.get("tasa_pactada")),
"reestructurado": to_categorical(raw.get("reestructurado")),
"status_credito": to_categorical(raw.get("status_credito")),
"metodo_calculo": to_categorical(raw.get("metodo_calculo")),
"tipo_garantia": to_categorical(raw.get("tipo_garantia")),
"frecuencia_pago": to_categorical(raw.get("frecuencia_pago")),
"id_agencia": to_categorical(raw.get("id_agencia")),
"tipo_cliente": to_categorical(raw.get("tipo_cliente")),
"categoria_cliente": to_categorical(raw.get("categoria_cliente")),
"persona_pep": to_categorical(raw.get("persona_pep")),
"persona_cpe": to_categorical(raw.get("persona_cpe")),
**{key: value for key, value in derived.items() if key != "fecha_snapshot"},
}
return feature_map, derived
def get_categorical_feature_indices(schema: Mapping[str, Any]) -> list[int]:
categorical = set(schema["categorical_features"])
return [
index
for index, feature_name in enumerate(schema["feature_order"])
if feature_name in categorical
]
def build_catboost_row(
raw: Mapping[str, Any],
schema: Mapping[str, Any],
) -> tuple[list[Any], dict[str, Any], dict[str, Any]]:
feature_map, derived = build_feature_map(raw)
categorical = set(schema["categorical_features"])
row: list[Any] = []
for feature_name in schema["feature_order"]:
value = feature_map.get(feature_name)
if feature_name in categorical:
row.append(to_categorical(value))
else:
row.append(float("nan") if value is None else float(value))
return row, feature_map, derived
|