{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "177faf3e-cec4-4a9f-a613-4c8838af30c1",
   "metadata": {},
   "outputs": [],
   "source": [
    "#No keras existem os modelos de redes neurais estaticos, funcionais e dinamicmos. Neste notebook lidaresmos com dinamico"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ef8dbd4c-3443-4a5f-b8e1-99b6f0d958b6",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.datasets import fetch_california_housing\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "import pandas as pd\n",
    "import matplotlib\n",
    "matplotlib.use(\"TkAgg\")\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "09438fbc-8705-4712-8111-078a22cf8d1d",
   "metadata": {},
   "outputs": [],
   "source": [
    "housing = fetch_california_housing()\n",
    "x_train_full, x_test, y_train_full, y_test = train_test_split(housing.data,housing.target)\n",
    "x_train, x_valid, y_train, y_valid = train_test_split(x_train_full, y_train_full)\n",
    "\n",
    "scaler=StandardScaler()\n",
    "x_train = scaler.fit_transform(x_train)\n",
    "x_valid = scaler.transform(x_valid)\n",
    "x_test=scaler.transform(x_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0e80a374-8f22-40cd-ad12-f15898142f6b",
   "metadata": {},
   "outputs": [],
   "source": [
    "class WideAndDeepModel(keras.Model):\n",
    "    def __init__(self, units=30, activation=\"relu\", **kwargs):\n",
    "        super().__init__(**kwargs)  \n",
    "        self.hidden1 = keras.layers.Dense(units, activation=activation)\n",
    "        self.hidden2 = keras.layers.Dense(units, activation=activation)\n",
    "        self.main_output = keras.layers.Dense(1)\n",
    "        self.aux_output = keras.layers.Dense(1)\n",
    "    def call(self, inputs):\n",
    "        input_A, input_B = inputs\n",
    "        hidden1 = self.hidden1(input_B)\n",
    "        hidden2 = self.hidden2(hidden1)\n",
    "        concat = keras.layers.concatenate([input_A, hidden2])\n",
    "        main_output = self.main_output(concat)\n",
    "        aux_output = self.aux_output(hidden2)\n",
    "        return main_output, aux_output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3a81c2b8-4e95-453a-8d1a-4ba132bd483b",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = WideAndDeepModel()\n",
    "\n",
    "model.compile(\n",
    "    loss=[\"mse\", \"mse\"],           # uma loss para cada saída\n",
    "    optimizer=keras.optimizers.SGD(learning_rate=1e-3),\n",
    "    loss_weights=[0.8, 0.2]        # opcional: peso de cada saída\n",
    ")\n",
    "\n",
    "# Os targets precisam ser tuplas/listas correspondentes às saídas\n",
    "history = model.fit(\n",
    "    (x_train_A, x_train_B),        # entradas\n",
    "    (y_train, y_train),            # targets: main_output e aux_output\n",
    "    epochs=20,\n",
    "    validation_data=((x_valid_A, x_valid_B), (y_valid, y_valid))\n",
    ")\n",
    "\n",
    "# Avaliar\n",
    "mse_test = model.evaluate((x_test_A, x_test_B), (y_test, y_test))\n",
    "\n",
    "# Predição\n",
    "y_pred_main, y_pred_aux = model.predict((x_new_A, x_new_B))\n",
    "print(y_pred_main)\n",
    "\n",
    "model.save(\"my_keras_model.h5\") #Salvando modelo, legal isso"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "044977e1-310c-4f39-9e46-303c86bfbd0f",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = keras.models.load_model(\"my_keras_model.h5\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "079566ef-1c0c-400a-aeba-253ba01d1d5c",
   "metadata": {},
   "outputs": [],
   "source": [
    "checkpoint_cb = keras.callbacs.ModelCheckpoint(\"my_keras_model.h5\")\n",
    "history = model.fit(x_train, y_train, epochs = 10, callbacks = [checkpoint_cb])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8b95335e-9efc-44d9-8b6f-9cf8c6929dc7",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.12.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}