Upload notebooks/02_inferencia.ipynb with huggingface_hub

notebooks/02_inferencia.ipynb

@@ -0,0 +1,146 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "cell30",
+   "metadata": {},
+   "source": [
+    "[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/#fileId=https%3A%2F%2Fhuggingface.co%2Fgusdelact%2Fbond-trade-price-histgb%2Fresolve%2Fmain%2Fnotebooks%2F02_inferencia.ipynb)\n",
+    "\n",
+    "# 02 \u2014 Inferencia (Bond Trade Price)\n",
+    "\n",
+    "Descarga el modelo entrenado desde HF Hub y predice `trade_price`. No reentrena nada.\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "id": "cell31",
+   "metadata": {},
+   "execution_count": null,
+   "outputs": [],
+   "source": [
+    "import importlib, subprocess\n",
+    "def _ensure(pkg, import_name=None):\n",
+    "    name = import_name or pkg.split('[')[0].split('==')[0].split('>=')[0]\n",
+    "    try:\n",
+    "        importlib.import_module(name)\n",
+    "    except ImportError:\n",
+    "        subprocess.check_call(['uv','pip','install','--system','--quiet',pkg])\n",
+    "_ensure('huggingface-hub>=0.28.1', 'huggingface_hub')\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "id": "cell32",
+   "metadata": {},
+   "execution_count": null,
+   "outputs": [],
+   "source": [
+    "import json, joblib, sklearn\n",
+    "from pathlib import Path\n",
+    "import numpy as np, pandas as pd\n",
+    "from huggingface_hub import hf_hub_download\n",
+    "\n",
+    "REPO_ID = 'gusdelact/bond-trade-price-histgb'\n",
+    "model = joblib.load(hf_hub_download(REPO_ID, 'model.joblib'))\n",
+    "info  = json.loads(Path(hf_hub_download(REPO_ID, 'model_info.json')).read_text())\n",
+    "exp = info['library_versions'].get('scikit-learn')\n",
+    "if exp and exp != sklearn.__version__:\n",
+    "    print(f'[warn] sklearn instalado={sklearn.__version__}, modelo={exp}. Si falla, reentrena.')\n",
+    "FEATURE_ORDER = info['feature_order']; DEFAULTS = info['defaults']\n",
+    "print('modelo:', info['model_name'], '| MAE test:', info['test_metrics']['mae'] if 'test_metrics' in info else info.get('metrics'))\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "cell33",
+   "metadata": {},
+   "source": [
+    "## Funcion de prediccion\n",
+    "\n",
+    "Envuelve el input en un DataFrame con las columnas en el orden de `feature_order` (evita warnings y cruces silenciosos). Las features no provistas usan `defaults`."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "id": "cell34",
+   "metadata": {},
+   "execution_count": null,
+   "outputs": [],
+   "source": [
+    "def predict(input_dict: dict) -> float:\n",
+    "    row = dict(DEFAULTS)\n",
+    "    row.update({k: v for k, v in input_dict.items() if k in FEATURE_ORDER})\n",
+    "    X = pd.DataFrame([{c: row[c] for c in FEATURE_ORDER}])\n",
+    "    return float(np.clip(model.predict(X), 0, None)[0])\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "cell35",
+   "metadata": {},
+   "source": [
+    "## Ejemplo feliz"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "id": "cell36",
+   "metadata": {},
+   "execution_count": null,
+   "outputs": [],
+   "source": [
+    "ejemplo = {'curve_based_price': 101.5, 'trade_price_last1': 101.2,\n",
+    "           'current_coupon': 5.0, 'time_to_maturity': 8.0, 'trade_size': 100000,\n",
+    "           'trade_type': 2}\n",
+    "print('precio estimado:', round(predict(ejemplo), 3))\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "cell37",
+   "metadata": {},
+   "source": [
+    "## Ejemplo de error (feature desconocida se ignora; faltantes usan default)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "id": "cell38",
+   "metadata": {},
+   "execution_count": null,
+   "outputs": [],
+   "source": [
+    "try:\n",
+    "    print('con feature inexistente:', round(predict({'foo_bar': 1.0}), 3),\n",
+    "          '(usa todos los defaults)')\n",
+    "except Exception as e:\n",
+    "    print('error:', e)\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "cell39",
+   "metadata": {},
+   "source": [
+    "## Bonus \u2014 comparacion con el HF Space desplegado (opcional)\n",
+    "\n",
+    "El modelo esta desplegado en https://huggingface.co/spaces/gusdelact/bond-trade-price-predictor\n",
+    "La logica de `predict` aqui coincide con la de `app_inference/app_inference.py`."
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "name": "python",
+   "version": "3.12"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}