diff --git "a/S\303\251ries temporelles/Reseau_RNN.ipynb" "b/S\303\251ries temporelles/Reseau_RNN.ipynb"
index 384cc98..d3438ff 100644
--- "a/S\303\251ries temporelles/Reseau_RNN.ipynb"	
+++ "b/S\303\251ries temporelles/Reseau_RNN.ipynb"	
@@ -1,23 +1,9 @@
 {
-  "nbformat": 4,
-  "nbformat_minor": 0,
-  "metadata": {
-    "colab": {
-      "name": "Reseau_RNN.ipynb",
-      "provenance": [],
-      "include_colab_link": true
-    },
-    "kernelspec": {
-      "name": "python3",
-      "display_name": "Python 3"
-    }
-  },
   "cells": [
     {
       "cell_type": "markdown",
       "metadata": {
-        "id": "view-in-github",
-        "colab_type": "text"
+        "id": "view-in-github"
       },
       "source": [
         "<a href=\"https://colab.research.google.com/github/AlexandreBourrieau/ML2/blob/main/S%C3%A9ries%20temporelles/Reseau_RNN.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
@@ -35,18 +21,18 @@
     },
     {
       "cell_type": "code",
+      "execution_count": null,
       "metadata": {
         "id": "RRhtHsNn5fc3"
       },
+      "outputs": [],
       "source": [
         "import tensorflow as tf\n",
         "from tensorflow import keras\n",
         "\n",
         "import numpy as np\n",
         "import matplotlib.pyplot as plt"
-      ],
-      "execution_count": 1,
-      "outputs": []
+      ]
     },
     {
       "cell_type": "markdown",
@@ -59,9 +45,11 @@
     },
     {
       "cell_type": "code",
+      "execution_count": null,
       "metadata": {
         "id": "vJfSLtub6sdc"
       },
+      "outputs": [],
       "source": [
         "# Fonction permettant d'afficher une série temporelle\n",
         "def affiche_serie(temps, serie, format=\"-\", debut=0, fin=None, label=None):\n",
@@ -105,9 +93,7 @@
         "affiche_serie(temps,serie)\n",
         "plt.title('Série temporelle expérimentale')\n",
         "plt.show()"
-      ],
-      "execution_count": null,
-      "outputs": []
+      ]
     },
     {
       "cell_type": "markdown",
@@ -120,9 +106,11 @@
     },
     {
       "cell_type": "code",
+      "execution_count": null,
       "metadata": {
         "id": "78vEUlpIFMp6"
       },
+      "outputs": [],
       "source": [
         "# Fonction permettant de créer un dataset à partir des données de la série temporelle\n",
         "# au format X(X1,X2,...Xn) / Y(Y1,Y2,...,Yn)\n",
@@ -136,9 +124,7 @@
         "  dataset = dataset.shuffle(buffer_melange).map(lambda x: (x[:-1], x[-1:]))\n",
         "  dataset = dataset.batch(batch_size, drop_remainder=True).prefetch(1)\n",
         "  return dataset"
-      ],
-      "execution_count": null,
-      "outputs": []
+      ]
     },
     {
       "cell_type": "markdown",
@@ -160,9 +146,11 @@
     },
     {
       "cell_type": "code",
+      "execution_count": null,
       "metadata": {
         "id": "_14USRg5JvSu"
       },
+      "outputs": [],
       "source": [
         "temps_separation = 1000\n",
         "\n",
@@ -173,9 +161,7 @@
         "# Exctraction des temps et des données de valiadation\n",
         "temps_validation = temps[temps_separation:]\n",
         "x_validation = serie[temps_separation:]"
-      ],
-      "execution_count": null,
-      "outputs": []
+      ]
     },
     {
       "cell_type": "markdown",
@@ -201,14 +187,16 @@
         "id": "C2CDLaYDoDms"
       },
       "source": [
-        "<img src=\"https://github.com/AlexandreBourrieau/ML/blob/main/Carnets%20Jupyter/S%C3%A9ries%20temporelles/images/split_XY_2.png?raw=true\" width=\"1200\"> "
+        "<img src=\"https://github.com/AlexandreBourrieau/ML/blob/main/Carnets%20Jupyter/S%C3%A9ries%20temporelles/images/split_XY_2.png?raw=true\" width=\"1200\">"
       ]
     },
     {
       "cell_type": "code",
+      "execution_count": null,
       "metadata": {
         "id": "ziajfOefKvsu"
       },
+      "outputs": [],
       "source": [
         "# Définition des caractéristiques du dataset que l'on souhaite créer\n",
         "taille_fenetre = 20\n",
@@ -220,9 +208,7 @@
         "\n",
         "# Création du dataset X,Y de validation\n",
         "dataset_Val = prepare_dataset_XY(x_validation,taille_fenetre,batch_size,buffer_melange)"
-      ],
-      "execution_count": null,
-      "outputs": []
+      ]
     },
     {
       "cell_type": "markdown",
@@ -302,20 +288,20 @@
         "id": "1OZkfsmnBNHY"
       },
       "source": [
-        "<img src=\"https://github.com/AlexandreBourrieau/ML/blob/main/Carnets%20Jupyter/S%C3%A9ries%20temporelles/images/RNN_.png?raw=true\" width=\"1200\"> "
+        "<img src=\"https://github.com/AlexandreBourrieau/ML/blob/main/Carnets%20Jupyter/S%C3%A9ries%20temporelles/images/RNN_.png?raw=true\" width=\"1200\">"
       ]
     },
     {
       "cell_type": "code",
+      "execution_count": null,
       "metadata": {
         "id": "5zOZ3yVyZmkl"
       },
+      "outputs": [],
       "source": [
         "# Remise à zéro de tous les états générés par Keras\n",
         "tf.keras.backend.clear_session()"
-      ],
-      "execution_count": null,
-      "outputs": []
+      ]
     },
     {
       "cell_type": "markdown",
@@ -323,14 +309,16 @@
         "id": "86TUv-QxEFy9"
       },
       "source": [
-        "Pour insérer une dimension de type `None` au format de l'entrée, on utilise la méthode [expand_dims](https://www.tensorflow.org/api_docs/python/tf/expand_dims) de tensorflow. "
+        "Pour insérer une dimension de type `None` au format de l'entrée, on utilise la méthode [expand_dims](https://www.tensorflow.org/api_docs/python/tf/expand_dims) de tensorflow."
       ]
     },
     {
       "cell_type": "code",
+      "execution_count": null,
       "metadata": {
         "id": "LORqSH0DQKoG"
       },
+      "outputs": [],
       "source": [
         "# Fonction de la couche lambda d'entrée\n",
         "def Traitement_Entrees(x):\n",
@@ -357,7 +345,48 @@
         "model.save_weights(\"model_initial.hdf5\")\n",
         "\n",
         "model.summary()"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "source": [
+        "# Normalisation des données"
       ],
+      "metadata": {
+        "id": "zJ0oY-QBxfqL"
+      }
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "# Calcul de la moyenne et de l'écart type de la série\n",
+        "mean = tf.math.reduce_mean(np.asarray(serie))\n",
+        "std = tf.math.reduce_std(np.asarray((serie)))\n",
+        "\n",
+        "# Normalisation des données\n",
+        "Serie_Normalisee = (serie-mean)/std\n",
+        "serie_test = (serie-mean)/std"
+      ],
+      "metadata": {
+        "id": "pa6ULOIAyNJ_"
+      },
+      "execution_count": null,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "# Extraction des temps et des données d'entrainement\n",
+        "x_entrainement_norm = Serie_Normalisee[:temps_separation]\n",
+        "x_validation_norm = Serie_Normalisee[temps_separation:]\n",
+        "\n",
+        "dataset_norm = prepare_dataset_XY(x_entrainement_norm, taille_fenetre, batch_size, buffer_melange)\n",
+        "\n",
+        "dataset_Val_norm = prepare_dataset_XY(x_validation_norm, taille_fenetre, batch_size, buffer_melange)"
+      ],
+      "metadata": {
+        "id": "_XCRrtuZyjqs"
+      },
       "execution_count": null,
       "outputs": []
     },
@@ -385,7 +414,7 @@
         "id": "J-cR2CjAVg6j"
       },
       "source": [
-        "<img src=\"https://github.com/AlexandreBourrieau/ML/blob/main/Carnets%20Jupyter/S%C3%A9ries%20temporelles/images/ModifGene.png?raw=true\" width=\"600\"> "
+        "<img src=\"https://github.com/AlexandreBourrieau/ML/blob/main/Carnets%20Jupyter/S%C3%A9ries%20temporelles/images/ModifGene.png?raw=true\" width=\"600\">"
       ]
     },
     {
@@ -412,27 +441,29 @@
         "id": "XYL9uZn4Vtyb"
       },
       "source": [
-        "<img src=\"https://github.com/AlexandreBourrieau/ML/blob/main/Carnets%20Jupyter/S%C3%A9ries%20temporelles/images/ModifGene_2.png?raw=true\" width=\"1200\"> "
+        "<img src=\"https://github.com/AlexandreBourrieau/ML/blob/main/Carnets%20Jupyter/S%C3%A9ries%20temporelles/images/ModifGene_2.png?raw=true\" width=\"1200\">"
       ]
     },
     {
       "cell_type": "code",
+      "execution_count": null,
       "metadata": {
         "id": "dqGYeKqJOlua"
       },
+      "outputs": [],
       "source": [
         "# Remise à zéro des états du modèle\n",
         "tf.keras.backend.clear_session()\n",
         "model.load_weights(\"model_initial.hdf5\")"
-      ],
-      "execution_count": null,
-      "outputs": []
+      ]
     },
     {
       "cell_type": "code",
+      "execution_count": null,
       "metadata": {
         "id": "9sLSLTvjOzCM"
       },
+      "outputs": [],
       "source": [
         "# Fonction de la couche lambda d'entrée\n",
         "def Traitement_Entrees(x):\n",
@@ -455,9 +486,7 @@
         "# Construction du modèle\n",
         "model = tf.keras.Model(entrees,sortie)\n",
         "model.summary()"
-      ],
-      "execution_count": null,
-      "outputs": []
+      ]
     },
     {
       "cell_type": "markdown",
@@ -470,16 +499,18 @@
     },
     {
       "cell_type": "code",
+      "execution_count": null,
       "metadata": {
         "id": "2uU0jcwFOlub"
       },
+      "outputs": [],
       "source": [
         "# Définition de la fonction de régulation du taux d'apprentissage\n",
         "def RegulationTauxApprentissage(periode, taux):\n",
         "  return 1e-8*20**(periode/10)\n",
         "\n",
         "# Définition de l'optimiseur à utiliser\n",
-        "optimiseur=tf.keras.optimizers.SGD(lr=1e-8, momentum=0.9)\n",
+        "optimiseur=tf.keras.optimizers.SGD(lr=6e-4, momentum=0.9)\n",
         "\n",
         "# Utilisation de la méthode ModelCheckPoint\n",
         "CheckPoint = tf.keras.callbacks.ModelCheckpoint(\"poids.hdf5\", monitor='loss', verbose=1, save_best_only=True, save_weights_only = True, mode='auto', save_freq='epoch')\n",
@@ -489,17 +520,17 @@
         "\n",
         "# Entraine le modèle en utilisant la méthode ModelCheckpoint pour sauvegarder les meilleurs poids du modèle\n",
         "historique = model.fit(dataset_norm, epochs=100,verbose=1, callbacks=[tf.keras.callbacks.LearningRateScheduler(RegulationTauxApprentissage), CheckPoint])"
-      ],
-      "execution_count": null,
-      "outputs": []
+      ]
     },
     {
       "cell_type": "code",
+      "execution_count": null,
       "metadata": {
         "id": "9iMyNGyXOlub"
       },
+      "outputs": [],
       "source": [
-        "# Construit un vecteur avec les valeurs du taux d'apprentissage à chaque période \n",
+        "# Construit un vecteur avec les valeurs du taux d'apprentissage à chaque période\n",
         "taux = 1e-8*(10**(np.arange(100)/10))\n",
         "\n",
         "# Affiche l'erreur en fonction du taux d'apprentissage\n",
@@ -507,27 +538,27 @@
         "plt.semilogx(taux,historique.history[\"loss\"])\n",
         "plt.axis([ 1e-8, 1e-1, 0, 0.2])\n",
         "plt.title(\"Evolution de l'erreur en fonction du taux d'apprentissage\")"
-      ],
-      "execution_count": null,
-      "outputs": []
+      ]
     },
     {
       "cell_type": "code",
+      "execution_count": null,
       "metadata": {
         "id": "QAIF0Gy2Oluc"
       },
+      "outputs": [],
       "source": [
         "# Chargement des poids sauvegardés\n",
         "model.load_weights(\"poids.hdf5\")"
-      ],
-      "execution_count": null,
-      "outputs": []
+      ]
     },
     {
       "cell_type": "code",
+      "execution_count": null,
       "metadata": {
         "id": "crC29nqwOluc"
       },
+      "outputs": [],
       "source": [
         "from timeit import default_timer as timer\n",
         "\n",
@@ -567,15 +598,15 @@
         "infos = cb.GetInfos()\n",
         "print(\"Step time : %.3f\" %infos[0])\n",
         "print(\"Total time : %.3f\" %infos[1])"
-      ],
-      "execution_count": null,
-      "outputs": []
+      ]
     },
     {
       "cell_type": "code",
+      "execution_count": null,
       "metadata": {
         "id": "a2g48gUlOlud"
       },
+      "outputs": [],
       "source": [
         "erreur_entrainement = historique.history[\"loss\"]\n",
         "erreur_validation = historique.history[\"val_loss\"]\n",
@@ -587,15 +618,15 @@
         "plt.legend()\n",
         "\n",
         "plt.title(\"Evolution de l'erreur en fonction de la période\")"
-      ],
-      "execution_count": null,
-      "outputs": []
+      ]
     },
     {
       "cell_type": "code",
+      "execution_count": null,
       "metadata": {
         "id": "ZZaLzXnpOlud"
       },
+      "outputs": [],
       "source": [
         "erreur_entrainement = historique.history[\"loss\"]\n",
         "erreur_validation = historique.history[\"val_loss\"]\n",
@@ -607,9 +638,7 @@
         "plt.legend()\n",
         "\n",
         "plt.title(\"Evolution de l'erreur en fonction de la période\")"
-      ],
-      "execution_count": null,
-      "outputs": []
+      ]
     },
     {
       "cell_type": "markdown",
@@ -622,9 +651,11 @@
     },
     {
       "cell_type": "code",
+      "execution_count": null,
       "metadata": {
         "id": "_h5wZ20yOlud"
       },
+      "outputs": [],
       "source": [
         "taille_fenetre = 20\n",
         "\n",
@@ -636,15 +667,15 @@
         "for t in temps[temps_separation:-taille_fenetre]:\n",
         "    X = np.reshape(Serie_Normalisee[t:t+taille_fenetre],(1,taille_fenetre))\n",
         "    predictions.append(model.predict(X))"
-      ],
-      "execution_count": null,
-      "outputs": []
+      ]
     },
     {
       "cell_type": "code",
+      "execution_count": null,
       "metadata": {
         "id": "hLP355JJOlue"
       },
+      "outputs": [],
       "source": [
         "# Affiche la série et les prédictions\n",
         "plt.figure(figsize=(10, 6))\n",
@@ -659,26 +690,35 @@
         "affiche_serie(temps[temps_separation+taille_fenetre:],np.asarray((predictions*std+mean))[:,0,0],label=\"Prédictions\")\n",
         "plt.title(\"Prédictions avec le réseau récurrent (zoom sur l'intervalle de validation)\")\n",
         "plt.show()"
-      ],
-      "execution_count": null,
-      "outputs": []
+      ]
     },
     {
       "cell_type": "code",
+      "execution_count": null,
       "metadata": {
         "id": "4WkrGXlcOlue"
       },
+      "outputs": [],
       "source": [
-        "# Calcule de l'erreur quadratique moyenne et de l'erreur absolue moyenne \n",
+        "# Calcule de l'erreur quadratique moyenne et de l'erreur absolue moyenne\n",
         "\n",
         "mae = tf.keras.metrics.mean_absolute_error(serie[temps_separation+taille_fenetre:],np.asarray((predictions*std+mean))[:,0,0]).numpy()\n",
         "mse = tf.keras.metrics.mean_squared_error(serie[temps_separation+taille_fenetre:],np.asarray((predictions*std+mean))[:,0,0]).numpy()\n",
         "\n",
         "print(mae)\n",
         "print(mse)"
-      ],
-      "execution_count": null,
-      "outputs": []
+      ]
     }
-  ]
-}
\ No newline at end of file
+  ],
+  "metadata": {
+    "colab": {
+      "provenance": []
+    },
+    "kernelspec": {
+      "display_name": "Python 3",
+      "name": "python3"
+    }
+  },
+  "nbformat": 4,
+  "nbformat_minor": 0
+}