An application of machine learning for the diagnosis of breast cancer.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# An application of machine learning for the diagnosis of breast cancer\n",
    "\n",
    "Welcome to our first turorial in jupyter, in which you will see the analysis of breast cancer data extracted from the sklearn.datasetslibrary,through different type of modeling, such as:\n",
    "\n",
    "- decision tree \n",
    "- nayves bayes\n",
    "- SVM (Support vector machine)\n",
    "- Random forest\n",
    "\n",
    "### 1 - Packages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Load the library with the breast cancer dataset\n",
    "from sklearn.datasets import load_breast_cancer\n",
    "\n",
    "# Load pandas\n",
    "import pandas as pd\n",
    "# Load numpy\n",
    "import numpy as np\n",
    "\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2 - Overview of the data set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Asiganar un random seed\n",
    "np.random.seed(0)\n",
    "# Creas un objeto llamado cancer \n",
    "cancer=load_breast_cancer()\n",
    "# Creas una dataframe con 30 feature variables\n",
    "df = pd.DataFrame(cancer.data, columns=cancer.feature_names)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3 - Separate the data "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of observations in the training data: 435\n",
      "Number of observations in the test data: 134\n"
     ]
    }
   ],
   "source": [
    "# Agregamos una nueva columna con the species names,Esto es lo que intentaremos predecir\n",
    "df['species'] = pd.Categorical.from_codes(cancer.target, cancer.target_names)\n",
    "\n",
    "\n",
    "# separamos la data.\n",
    "df['is_train'] = np.random.uniform(0, 1, len(df)) <= .75\n",
    "\n",
    "\n",
    "\n",
    "# Creamos 2 nuevas dataframe , una con the training rows, otra con the test rows\n",
    "train, test = df[df['is_train']==True], df[df['is_train']==False]\n",
    "\n",
    "\n",
    "# Mostramos los nuemros de observaciones para the test y training dataframes\n",
    "print('Number of observations in the training data:', len(train))\n",
    "print('Number of observations in the test data:',len(test))\n",
    "\n",
    "# Creamos una lista de the feature column's names\n",
    "features = df.columns[:30]      \n",
    "\n",
    "\n",
    "y = pd.factorize(train['species'])[0]\n",
    "y_test = pd.factorize(test['species'])[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4 - Using a decision tree "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 4.1 - Import tree classifier library"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "#from sklearn.ensemble import tree classifier library\n",
    "from sklearn import tree"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 4.2 - Create a DecisionTree and train it"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9104477611940298\n",
      "['malignant' 'malignant' 'malignant' 'malignant' 'malignant']\n",
      "7     malignant\n",
      "8     malignant\n",
      "10    malignant\n",
      "13    malignant\n",
      "17    malignant\n",
      "Name: species, dtype: category\n",
      "Categories (2, object): [malignant, benign]\n"
     ]
    }
   ],
   "source": [
    "# Create a DecisionTree\n",
    "\n",
    "clf=tree.DecisionTreeClassifier()\n",
    "# Train the Classifier to take the training features and learn how they relate\n",
    "# to the training y (the species)\n",
    "clf.fit(train[features], y)\n",
    "\n",
    "predic=clf.predict(test[features])\n",
    "error=clf.score(test[features],y_test,sample_weight=None)\n",
    "print(error)\n",
    "#print(clf.predict(test[features]))\n",
    "\n",
    "predict_porcentaje=clf.predict_proba(test[features])[0:50] \n",
    "\n",
    "#print(clf.predict_proba(test[features])[0:50] )\n",
    "\n",
    "preds = cancer.target_names[clf.predict(test[features])]\n",
    "\n",
    "print(preds[0:5])\n",
    "\n",
    "print(test['species'].head())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 4.3 - Create confusion matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predicted Species  benign  malignant\n",
      "Actual Species                      \n",
      "malignant               5         43\n",
      "benign                 79          7\n"
     ]
    }
   ],
   "source": [
    "\n",
    "matrix=pd.crosstab(test['species'], preds, rownames=['Actual Species'], colnames=['Predicted Species'])\n",
    "\n",
    "print(matrix)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5 -  Using a naive bayes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 5.1 - Import naive bayes library"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "#from sklearn import naive bayes\n",
    "from sklearn.naive_bayes import GaussianNB\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 5.2 - Create a naive bayes and train it"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9402985074626866\n",
      "['malignant' 'malignant' 'malignant' 'benign' 'malignant']\n",
      "7     malignant\n",
      "8     malignant\n",
      "10    malignant\n",
      "13    malignant\n",
      "17    malignant\n",
      "Name: species, dtype: category\n",
      "Categories (2, object): [malignant, benign]\n"
     ]
    }
   ],
   "source": [
    "#Creamos un Gaussiano\n",
    "clf=GaussianNB()\n",
    "\n",
    "# Train the Classifier to take the training features and learn how they relate\n",
    "# to the training y (the species)\n",
    "clf.fit(train[features], y)\n",
    "\n",
    "predic=clf.predict(test[features])\n",
    "error=clf.score(test[features],y_test,sample_weight=None)\n",
    "print(error)\n",
    "#print(clf.predict(test[features]))\n",
    "\n",
    "predict_porcentaje=clf.predict_proba(test[features])[0:50] \n",
    "\n",
    "#print(clf.predict_proba(test[features])[0:50] )\n",
    "\n",
    "preds = cancer.target_names[clf.predict(test[features])]\n",
    "\n",
    "print(preds[0:5])\n",
    "\n",
    "print(test['species'].head())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 5.3 - Create confusion matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predicted Species  benign  malignant\n",
      "Actual Species                      \n",
      "malignant               5         43\n",
      "benign                 83          3\n"
     ]
    }
   ],
   "source": [
    " # Create confusion matrix\n",
    "matrix=pd.crosstab(test['species'], preds, rownames=['Actual Species'], colnames=['Predicted Species'])\n",
    "print(matrix)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 6 -  Using a Support Vector Machine"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 6.1 - Import  SVM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "#De sklearn importamos svm\n",
    "from sklearn import svm\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 6.2 - Create a SVM and train it"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9626865671641791\n",
      "['malignant' 'malignant' 'malignant' 'benign' 'malignant']\n",
      "7     malignant\n",
      "8     malignant\n",
      "10    malignant\n",
      "13    malignant\n",
      "17    malignant\n",
      "Name: species, dtype: category\n",
      "Categories (2, object): [malignant, benign]\n"
     ]
    }
   ],
   "source": [
    "# Create a SVM\n",
    "clf=svm.SVC(C=1,kernel='linear')\n",
    "\n",
    "\n",
    "# Train the Classifier to take the training features and learn how they relate\n",
    "\n",
    "# PREDECIR SOLO 2 ATRIBUTOS\n",
    "#features=[\"worst perimeter\",\"mean concave points\"]\n",
    "clf.fit(train[features], y)\n",
    "\n",
    "predic=clf.predict(test[features])\n",
    "error=clf.score(test[features],y_test,sample_weight=None)\n",
    "print(error)\n",
    "\n",
    "\n",
    "preds = cancer.target_names[clf.predict(test[features])]\n",
    "\n",
    "print(preds[0:5])\n",
    "\n",
    "print(test['species'].head())\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 6.3 - Create confusion matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predicted Species  benign  malignant\n",
      "Actual Species                      \n",
      "malignant               5         43\n",
      "benign                 86          0\n"
     ]
    }
   ],
   "source": [
    "# Create confusion matrix\n",
    "matrix=pd.crosstab(test['species'], preds, rownames=['Actual Species'], colnames=['Predicted Species'])\n",
    "print(matrix)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 7 -  Using Random Forest"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 7.1 - Import  RandomForestClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "#from sklearn.ensemble import RandomForestClassifier \n",
    "from sklearn.ensemble import RandomForestClassifier"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 7.2 - Create a random forest Classifier and train it"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9552238805970149\n",
      "['malignant' 'malignant' 'malignant' 'malignant' 'malignant']\n",
      "7     malignant\n",
      "8     malignant\n",
      "10    malignant\n",
      "13    malignant\n",
      "17    malignant\n",
      "Name: species, dtype: category\n",
      "Categories (2, object): [malignant, benign]\n"
     ]
    }
   ],
   "source": [
    "# Creamos un random forest Classifier. By convention, clf means 'Classifier'\n",
    "clf = RandomForestClassifier(n_jobs=2, random_state=0,n_estimators=600)\n",
    "\n",
    "# Train the Classifier to take the training features and learn how they relate\n",
    "# to the training y (the species)\n",
    "\n",
    "#features=[\"worst perimeter\",\"mean concave points\"]\n",
    "clf.fit(train[features], y)\n",
    "\n",
    "predic=clf.predict(test[features])\n",
    "error=clf.score(test[features],y_test,sample_weight=None)\n",
    "print(error)\n",
    "#print(clf.predict(test[features]))\n",
    "\n",
    "predict_porcentaje=clf.predict_proba(test[features])[0:50] \n",
    "\n",
    "#print(clf.predict_proba(test[features])[0:50] )\n",
    "\n",
    "preds = cancer.target_names[clf.predict(test[features])]\n",
    "\n",
    "print(preds[0:5])\n",
    "\n",
    "print(test['species'].head())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 7.3 - Create confusion matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predicted Species  benign  malignant\n",
      "Actual Species                      \n",
      "malignant               4         44\n",
      "benign                 84          2\n"
     ]
    }
   ],
   "source": [
    "# Creamos una matrix de confusión\n",
    "matrix=pd.crosstab(test['species'], preds, rownames=['Actual Species'], colnames=['Predicted Species'])\n",
    "\n",
    "print(matrix)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see that the highest precision was obtained with the support vector machine\n",
    "\n",
    "Precision:\n",
    "\n",
    "- decision tree :0.91\n",
    "- nayves bayes :0.94\n",
    "- SVM (Support vector machine) :0.9626\n",
    "- Random forest :0.955\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 8 - Looking for important features\n",
    "\n",
    "After analyzing the precisions of each model, we look for which variables influence more when making the prediction."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x88d36b5d30>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "# View a list of the features and their importance scores\n",
    "importance= list(zip(train[features], clf.feature_importances_))\n",
    "\n",
    "#print(importance)\n",
    "\n",
    "n_features = cancer.data.shape[1]\n",
    "plt.barh(range(n_features), clf.feature_importances_, align='center')\n",
    "plt.yticks(np.arange(n_features), cancer.feature_names)\n",
    "plt.xlabel('Feature Importance')\n",
    "plt.ylabel('Feature')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 9 - Analyzing data with these 2 features and using SVM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x88d1594f98>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def make_meshgrid(x, y, h=.02):\n",
    "   \n",
    "    x_min, x_max = x.min() - 1, x.max() + 1\n",
    "    y_min, y_max = y.min() - 1, y.max() + 1\n",
    "    xx, yy = np.meshgrid(np.arange(x_min, x_max, h),\n",
    "                         np.arange(y_min, y_max, h))\n",
    "    return xx, yy\n",
    "\n",
    "\n",
    "def plot_contours(ax, clf, xx, yy, **params):\n",
    "    \n",
    "    Z = clf.predict(np.c_[xx.ravel(), yy.ravel()])\n",
    "    Z = Z.reshape(xx.shape)\n",
    "    out = ax.contourf(xx, yy, Z, **params)\n",
    "    return out\n",
    "\n",
    "import csv\n",
    " \n",
    "X = cancer.data[:,(7,22)]\n",
    "y = cancer.target\n",
    "\n",
    "# we create an instance of SVM and fit out data. We do not scale our\n",
    "# data since we want to plot the support vectors\n",
    "\n",
    "C = 1.0  # SVM regularization parameter\n",
    "models = (svm.SVC(kernel='linear', C=1),\n",
    "          svm.LinearSVC(C=C),\n",
    "          svm.SVC(kernel='rbf', gamma=0.7, C=1),\n",
    "          svm.SVC(kernel='poly', degree=3, C=1))\n",
    "models = (clf.fit(X, y) for clf in models)\n",
    "\n",
    "# title for the plots\n",
    "titles = ('SVC with linear kernel',\n",
    "          'LinearSVC (linear kernel)',\n",
    "          'SVC with RBF kernel',\n",
    "          'SVC with polynomial (degree 3) kernel')\n",
    "\n",
    "# Set-up 2x2 grid for plotting.\n",
    "fig, sub = plt.subplots(2, 2)\n",
    "plt.subplots_adjust(wspace=0.4, hspace=0.4)\n",
    "\n",
    "X0, X1 = X[:, 0], X[:, 1]\n",
    "xx, yy = make_meshgrid(X0, X1)\n",
    "\n",
    "for clf, title, ax in zip(models, titles, sub.flatten()):\n",
    "    plot_contours(ax, clf, xx, yy,\n",
    "                  cmap=plt.cm.coolwarm, alpha=0.8)\n",
    "    ax.scatter(X0, X1, c=y, cmap=plt.cm.coolwarm, s=20, edgecolors='k')\n",
    "    ax.set_xlim(xx.min(), xx.max())\n",
    "    ax.set_ylim(yy.min(), yy.max())\n",
    "    ax.set_xlabel(\"mean concave points\")\n",
    "    ax.set_ylabel(\"worst perimeter\")\n",
    "    ax.set_xticks(())\n",
    "    ax.set_yticks(())\n",
    "    ax.set_title(title)\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# THANKS !!!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "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": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}