Added comments

ann.ipynb

@@ -1,5 +1,13 @@
 {
  "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "4d162f18",
+   "metadata": {},
+   "source": [
+    "## Import du dataset"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 1,
@@ -23,6 +31,14 @@
     "from sklearn.neural_network import MLPClassifier"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "95e0ce45",
+   "metadata": {},
+   "source": [
+    "## Echantillonnage du dataset"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 4,
@@ -34,10 +50,15 @@
     "data = mnist.data.values[echantillon]\n",
     "target = mnist.target[echantillon]\n",
     "trainSize = 17500/25000\n",
-    "xtrain,  xtest,  ytrain,  ytest  =  train_test_split(data,  target, train_size=trainSize)\n",
+    "xtrain,  xtest,  ytrain,  ytest  =  train_test_split(data,  target, train_size=trainSize)"
-    "#trainSize = 49000/len(data)\n",
+   ]
-    "#xtrain,  xtest,  ytrain,  ytest  =  train_test_split(data,  target, \n",
+  },
-    "#train_size=trainSize)"
+  {
+   "cell_type": "markdown",
+   "id": "f35a2a10",
+   "metadata": {},
+   "source": [
+    "## Premier entraînement et scores"
    ]
   },
   {
@@ -110,6 +131,14 @@
     "precision_score(target.values, clf.predict(data), average='micro')"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "d5d7d4c1",
+   "metadata": {},
+   "source": [
+    "## Nombre de couches cachées optimal"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 14,
@@ -149,6 +178,14 @@
     "plt.show()"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "e4b13f21",
+   "metadata": {},
+   "source": [
+    "## Nombre de couches cachées et de neurones par couche optimaux"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 46,
@@ -231,6 +268,14 @@
     "plt.show()"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "11adbe03",
+   "metadata": {},
+   "source": [
+    "## Algorithme d'optimisation optimal"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 47,
@@ -370,6 +415,14 @@
     "plt.show()"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "cf7095a5",
+   "metadata": {},
+   "source": [
+    "## Fonction d'activation optimale"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 5,
@@ -517,6 +570,14 @@
     "plt.show()"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "569f6393",
+   "metadata": {},
+   "source": [
+    "## Valeur optimale d'alpha"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 62,
@@ -631,6 +692,14 @@
     "plt.show()"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "c46d042a",
+   "metadata": {},
+   "source": [
+    "## Matrice de confusion"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 6,

knn.ipynb

@@ -1,9 +1,17 @@
 {
  "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "87214e92",
+   "metadata": {},
+   "source": [
+    "## Import du dataset MNIST"
+   ]
+  },
   {
    "cell_type": "code",
-   "execution_count": 1,
+   "execution_count": null,
-   "id": "b4978985",
+   "id": "c2e6fa5e",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -11,6 +19,14 @@
     "mnist = fetch_openml('mnist_784') "
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "a36d996c",
+   "metadata": {},
+   "source": [
+    "## Manpulation du jeu de données"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 8,
@@ -703,6 +719,14 @@
     "print(mnist.target[4])"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "73194d9b",
+   "metadata": {},
+   "source": [
+    "## Premier entraînement"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 45,
@@ -747,6 +771,14 @@
     "clf.fit(xtrain, ytrain)"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "d7d80cd2",
+   "metadata": {},
+   "source": [
+    "## Vérification de la prédiction faite avec l'algorithme entraîné"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 68,
@@ -781,6 +813,14 @@
     "print(clf.predict([data[4]]))"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "8eb362f6",
+   "metadata": {},
+   "source": [
+    "## Scores obtenus"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 71,
@@ -805,6 +845,14 @@
     "print(clf.score(xtrain, ytrain))"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "aac45aac",
+   "metadata": {},
+   "source": [
+    "## Nombre de voisins optimal"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 3,
@@ -860,6 +908,14 @@
     "plt.show()"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "411f6841",
+   "metadata": {},
+   "source": [
+    "## Séparation optimale entre l'entraînement et le test "
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 18,
@@ -924,6 +980,14 @@
     "plt.show()"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "b3563ad5",
+   "metadata": {},
+   "source": [
+    "## Echantillonnage optimal du dataset"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 20,
@@ -985,6 +1049,14 @@
     "plt.show()"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "e387a386",
+   "metadata": {},
+   "source": [
+    "## Distance optimale"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 25,
@@ -1055,6 +1127,14 @@
     "plt.show()"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "c863ff41",
+   "metadata": {},
+   "source": [
+    "## Meilleure valeur de n_jobs "
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 31,
@@ -1129,6 +1209,14 @@
     "print(f'Temps d\\'exécution : {elapsed:.4}ms')"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "1a26da41",
+   "metadata": {},
+   "source": [
+    "## Matrice de confusion"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 7,

svm.ipynb

@@ -1,5 +1,13 @@
 {
  "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "557db042",
+   "metadata": {},
+   "source": [
+    "## Import du dataset"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 15,
@@ -11,6 +19,14 @@
     "mnist = fetch_openml('mnist_784')"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "2295fc5b",
+   "metadata": {},
+   "source": [
+    "## Premier entraînement et scores"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 16,
@@ -72,6 +88,14 @@
     "print(classifier.score(xtest, ytest))"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "32caafa6",
+   "metadata": {},
+   "source": [
+    "## Meilleur kernel à utiliser"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 7,
@@ -196,6 +220,8 @@
     }
    ],
    "source": [
+    "## Il est nécessaire de donner à l'algorithme en entrée une matrice carrée. \n",
+    "## Ne sachant pas comment faire nous avons décidé de ne pas prendre en compte ce type. \n",
     "start = time.time()\n",
     "clsvm = SVC(kernel='precomputed')\n",
     "clsvm.fit(xtrain, ytrain) \n",
@@ -243,6 +269,14 @@
     "plt.show()"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "218f2c18",
+   "metadata": {},
+   "source": [
+    "## Valeur optimale du paramètre C"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 8,
@@ -304,6 +338,14 @@
     "plt.show()"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "2ddd41c7",
+   "metadata": {},
+   "source": [
+    "## Matrice de confusion"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 19,
@@ -339,7 +381,7 @@
   {
    "cell_type": "code",
    "execution_count": 58,
-   "id": "14cd493b",
+   "id": "7ea957d9",
    "metadata": {},
    "outputs": [
     {