NeuralNetworksProject.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Neural Networks Project\n",
    "## Self-labeled techniques for semi-supervised learning\n",
    "<div style=\"text-align: right\">\n",
    "    Mark Laane <br />\n",
    "    Rome, 2017\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Introduction ###\n",
    "The aim of this project is to reimplement some techniques surveyed by Isaac Triguero et. al in paper [1] and to independently reproduce the reported results. A report of the project is also provided: [Project Report](Project Report Mark Laane.pdf)\n",
    "\n",
    "Two self-labeled techniques are chosen from the paper: Standard Self-Training and Tri-Training. Those techniques are used on Abalone and Dermatology datasets. For implementation, Python programming language was chosen along with Pandas and Sclearn libraries."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Datasets ###\n",
    "Two standard datasets are used: Abalone and Dermatology. The datasets are loaded from mldata.org"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#Datasets are stored in a python dictionary\n",
    "datasets = {}\n",
    "\n",
    "from sklearn.datasets import fetch_mldata\n",
    "from sklearn.pipeline import make_pipeline\n",
    "from sklearn.preprocessing import MinMaxScaler, OneHotEncoder\n",
    "\n",
    "# Fetch abalone dataset from mldata.org\n",
    "data = fetch_mldata(\"abalone\")\n",
    "# Preprocessing pipe for abalone dataset encodes categorical feature\n",
    "# and scales the features\n",
    "preprocessing_pipe = make_pipeline(\n",
    "    #OneHotEncoder on \"Sex\" feature\n",
    "    OneHotEncoder(categorical_features=[0], sparse=False),\n",
    "    #Scale all from 0 to 1\n",
    "    MinMaxScaler())\n",
    "# Apply preprocessing pipe to dataset and store the dataset in dict.\n",
    "datasets[\"abalone\"] = {\n",
    "    \"X\": preprocessing_pipe.fit_transform(data.data),\n",
    "    \"y\": data.target\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "#Dermatology dataset is loaded from mldata.org and used as-is\n",
    "data = fetch_mldata(\"uci-20070111 dermatology\")\n",
    "datasets[\"dermatology\"] = {\n",
    "    \"X\": data.data[:,0:-1],\n",
    "    \"y\": data.data[:,-1]\n",
    "}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Base classifiers ###\n",
    "3 different base classifiers are used. The base classifiers are provided by Sklearn library. The classifiers are configured according to the parameters described in the paper."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.naive_bayes import GaussianNB\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "# Class that only holds a collection of different \n",
    "# base classifiers for usage with SSL methods.\n",
    "class base_classifiers:\n",
    "    KNN = KNeighborsClassifier(\n",
    "        n_neighbors=3,\n",
    "        metric=\"euclidean\",\n",
    "        #n_jobs=2  # Parallelize work on CPUs\n",
    "    )\n",
    "    NB = GaussianNB(\n",
    "        priors=None\n",
    "    )\n",
    "    #SVM = SVC(\n",
    "    #    C=1.0,\n",
    "    #    kernel='poly',\n",
    "    #    degree=1,\n",
    "    #    tol=0.001,\n",
    "    #)\n",
    "    CART = DecisionTreeClassifier(\n",
    "        criterion='entropy',\n",
    "        # splitter='best',\n",
    "        # max_depth=None,\n",
    "        # min_samples_split=2,\n",
    "        min_samples_leaf=2,\n",
    "        # min_weight_fraction_leaf=0.0,\n",
    "        # max_features=None,\n",
    "        # random_state=None,\n",
    "        # max_leaf_nodes=None,\n",
    "        # min_impurity_split=1e-07,\n",
    "        # class_weight=None,\n",
    "        # presort=False,\n",
    "    )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Implemented self-labeled algorithms ###\n",
    "Two algorithms are implemented: Standard Self-Training and Tri Training.\n",
    "#### Standard Self-Training ####\n",
    "Implementation: [Standard Self-Training](standard_self_training.py)<br />\n",
    "The implementation is based on description of the algorithm in paper [2].\n",
    "Training an Standard Self-Training classifier is an iterative process - The base classifier is trained with initial labeled samples. Then it is used for labelling the unlabelled samples and the classifier is retrained with the most confident predictions. The process is repeated until the classifier output stabilizes.\n",
    "#### Tri-Training ###\n",
    "Implementation: [Tri-Training](tri_training.py)<br />\n",
    "The implementation is based on description of the algorithm in paper [3].\n",
    "In Tri-Training, Three base classifiers is trained on randomly subsampled sets of the labelled data. Then each of them will be iteratively trained on labeled data gained from two other base classifiers. The prediciton is made by using majority voting on three base classifiers.\n",
    "\n",
    "In total 6 different classifiers are trained - 2 techniques with 3 different base classifiers:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from standard_self_training import StandardSelfTraining\n",
    "from tri_training import TriTraining\n",
    "\n",
    "# All classifiers used for testing\n",
    "classifiers = [\n",
    "    TriTraining(\"TriTraining (KNN)\", base_classifiers.KNN),\n",
    "    TriTraining(\"TriTraining (NB)\", base_classifiers.NB),\n",
    "    #TriTraining(\"TriTraining (SVM)\", base_classifiers.SVM),\n",
    "    TriTraining(\"TriTraining (CART)\", base_classifiers.CART),\n",
    "    StandardSelfTraining(\"Self-Training (KNN)\", base_classifiers.KNN),\n",
    "    StandardSelfTraining(\"Self-Training (NB)\", base_classifiers.NB),\n",
    "    #StandardSelfTraining(\"Self-Training (SVM)\", base_classifiers.SVM),\n",
    "    StandardSelfTraining(\"Self-Training (CART)\", base_classifiers.CART)\n",
    "]\n",
    "labeling_rates = [0.10, 0.20, 0.30, 0.40]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Training and scoring ###"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import confusion_matrix\n",
    "import numpy as np\n",
    "\n",
    "def _training_scoring_iteration(clf, X, y, training_index, test_index, labeling_rate):\n",
    "    \"\"\" \n",
    "    One iteration of fully training and scoring a \n",
    "    classifier on given data (one Kfold split)\n",
    "    \"\"\"\n",
    "    #Testing set is set aside.. - 1/10th of the data\n",
    "    X_test, y_test = X[test_index], y[test_index]\n",
    "\n",
    "    #For generating a testing and transductive set\n",
    "    split_data = train_test_split(\n",
    "        X[training_index],\n",
    "        y[training_index],\n",
    "        test_size=labeling_rate,\n",
    "        random_state=42\n",
    "    )\n",
    "    (X_unlabeled, X_labeled, y_unlabeled, y_labeled) = split_data\n",
    "\n",
    "    #Training set - 9/10 of data\n",
    "    X_train = np.concatenate((X_labeled, X_unlabeled))\n",
    "    y_train = np.concatenate((\n",
    "        y_labeled.astype(str),\n",
    "        np.full_like(y_unlabeled.astype(str), \"unlabeled\")\n",
    "    ))\n",
    "    \n",
    "    #Train the classifier\n",
    "    clf.fit(X_train, y_train)\n",
    "    \n",
    "    #Score the classifier\n",
    "    transductive_score = clf.score(X_unlabeled, y_unlabeled.astype(str))\n",
    "    testing_score = clf.score(X_test, y_test.astype(str))\n",
    "\n",
    "    cnf_matrix = pd.DataFrame(\n",
    "        confusion_matrix(y_test.astype(str), clf.predict(X_test).astype(str))\n",
    "    )\n",
    "    \n",
    "    return transductive_score, testing_score, cnf_matrix\n",
    "    \n",
    "def train_and_score(clf, X, y, cv, labeling_rate):\n",
    "    \"\"\"\n",
    "    Perform KFold cross-validation of a classifier on a given data\n",
    "    and labelling rate\n",
    "    \"\"\"\n",
    "    transductive_scores = []\n",
    "    testing_scores = []\n",
    "    for training_index, test_index in cv.split(X,y):\n",
    "        transductive_score, testing_score, cnf_matrix = _training_scoring_iteration(clf, X, y, training_index, test_index, labeling_rate)\n",
    "        \n",
    "        transductive_scores.append(transductive_score)\n",
    "        testing_scores.append(testing_score)\n",
    "        print(\"#\", end=\"\")\n",
    "    print()\n",
    "    scores = {\n",
    "        \"trans_mean\": np.mean(transductive_scores),\n",
    "        \"test_mean\": np.mean(testing_scores),\n",
    "        \"trans_std\": np.std(transductive_scores),\n",
    "        \"test_std\": np.std(testing_scores)\n",
    "    }\n",
    "    return scores, cnf_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TriTraining (KNN)\n",
      "dataset: abalone \t\n",
      "rate: 0.1 ##"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "######"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#\n",
      "rate: 0.2 ######"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "###\n",
      "rate: 0.3 ###"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "####"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "##\n",
      "rate: 0.4 ###"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "###"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "###\n",
      "dataset: dermatology \t\n",
      "rate: 0.1 ##########\n",
      "rate: 0.2 ##########\n",
      "rate: 0.3 ##########\n",
      "rate: 0.4 ##########\n",
      "\n",
      "--------\n",
      "TriTraining (NB)\n",
      "dataset: abalone \t\n",
      "rate: 0.1 ##"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "##"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "###\n",
      "rate: 0.2 "
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "####"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "##"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#\n",
      "rate: 0.3 "
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "####"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "##"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "##\n",
      "rate: 0.4 "
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#####"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "###"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "##\n",
      "dataset: dermatology \t\n",
      "rate: 0.1 ##########\n",
      "rate: 0.2 ##########\n",
      "rate: 0.3 ##"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#####"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "###\n",
      "rate: 0.4 #####"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "##"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "###\n",
      "\n",
      "--------\n",
      "TriTraining (CART)\n",
      "dataset: abalone \t\n",
      "rate: 0.1 ###"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "######"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "#\n",
      "rate: 0.2 ########"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "##\n",
      "rate: 0.3 ##########\n",
      "rate: 0.4 ######"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/mark/.local/lib/python3.5/site-packages/numpy/ma/core.py:3883: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  check = self.filled(0).__eq__(other)\n",
      "/home/mark/.local/lib/python3.5/site-packages/sklearn/metrics/classification.py:177: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
      "  score = y_true == y_pred\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "####\n",
      "dataset: dermatology \t\n",
      "rate: 0.1 ##########\n",
      "rate: 0.2 ##########\n",
      "rate: 0.3 ##########\n",
      "rate: 0.4 ##########\n",
      "\n",
      "--------\n",
      "Self-Training (KNN)\n",
      "dataset: abalone \t\n",
      "rate: 0.1 35#28#27#30#26#30#36#29#29#36#\n",
      "rate: 0.2 30#39#29#24#27#33#34#29#31#31#\n",
      "rate: 0.3 25#26#25#26#22#21#23#27#30#28#\n",
      "rate: 0.4 23#20#19#19#20#22#24#21#21#21#\n",
      "dataset: dermatology \t\n",
      "rate: 0.1 16#10#10#15#12#11#13#11#15#14#\n",
      "rate: 0.2 19#14#10#15#11#10#12#9#8#17#\n",
      "rate: 0.3 14#13#10#10#10#8#11#9#10#8#\n",
      "rate: 0.4 7#8#8#10#11#10#9#7#9#8#\n",
      "\n",
      "--------\n",
      "Self-Training (NB)\n",
      "dataset: abalone \t\n",
      "rate: 0.1 19#20#10#12#9#22#19#13#29#16#\n",
      "rate: 0.2 9#15#28#29#33#37#40#40#37#40#\n",
      "rate: 0.3 39#20#37#40#40#40#40#40#40#40#\n",
      "rate: 0.4 40#40#40#40#40#40#40#40#40#40#\n",
      "dataset: dermatology \t\n",
      "rate: 0.1 1#1#1#3#1#1#1#1#26#28#\n",
      "rate: 0.2 2#40#40#40#40#40#40#40#40#36#\n",
      "rate: 0.3 40#40#40#40#40#40#40#40#40#40#\n",
      "rate: 0.4 40#40#40#40#40#40#40#40#40#40#\n",
      "\n",
      "--------\n",
      "Self-Training (CART)\n",
      "dataset: abalone \t\n",
      "rate: 0.1 19#21#19#21#16#12#25#19#15#20#\n",
      "rate: 0.2 23#23#27#20#21#31#22#25#26#18#\n",
      "rate: 0.3 19#26#26#24#22#24#27#21#18#19#\n",
      "rate: 0.4 21#21#22#23#24#20#20#24#24#26#\n",
      "dataset: dermatology \t\n",
      "rate: 0.1 1#1#1#1#1#1#3#2#2#2#\n",
      "rate: 0.2 1#1#2#1#2#3#3#4#3#3#\n",
      "rate: 0.3 3#3#2#5#4#4#4#6#3#2#\n",
      "rate: 0.4 1#1#3#2#2#3#3#3#5#3#\n",
      "\n",
      "--------\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import KFold\n",
    "import pandas as pd\n",
    "\n",
    "\"\"\" \n",
    "The main loop for testing \n",
    "all classifiers with \n",
    "all datasets and \n",
    "all labeling rates\n",
    "\"\"\"\n",
    "results = None\n",
    "cnf_matrixes = {}\n",
    "for classifier in classifiers:\n",
    "    cnf_matrixes[classifier.name] = {}\n",
    "    print(classifier.name)\n",
    "    for dataset_name, dataset in datasets.items():\n",
    "        cnf_matrixes[classifier.name][dataset_name] = {}\n",
    "        print(\"dataset:\", dataset_name, \"\\t\")\n",
    "        for labeling_rate in labeling_rates:\n",
    "            print(\"rate:\", labeling_rate, end=\" \")\n",
    "\n",
    "            test_info = { \"classifier\": classifier.name, \"dataset\":dataset_name, \"labeling_rate\":labeling_rate}\n",
    "            cv = KFold(n_splits=10, random_state=42)\n",
    "            scores, cnf_matrix = train_and_score(classifier, dataset[\"X\"], dataset[\"y\"], cv, labeling_rate)\n",
    "\n",
    "            if results is None:\n",
    "                results = pd.DataFrame([{**test_info, **scores}])\n",
    "            else:\n",
    "                results.loc[len(results.index)] = {**test_info, **scores}\n",
    "            cnf_matrixes[classifier.name][dataset_name][labeling_rate] = cnf_matrix\n",
    "    print()\n",
    "    print(\"--------\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Results ###"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>classifier</th>\n",
       "      <th>dataset</th>\n",
       "      <th>labeling_rate</th>\n",
       "      <th>test_mean</th>\n",
       "      <th>test_std</th>\n",
       "      <th>trans_mean</th>\n",
       "      <th>trans_std</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>TriTraining (KNN)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0.159610</td>\n",
       "      <td>0.093613</td>\n",
       "      <td>0.186119</td>\n",
       "      <td>0.063314</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>TriTraining (KNN)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.2</td>\n",
       "      <td>0.185034</td>\n",
       "      <td>0.084115</td>\n",
       "      <td>0.167961</td>\n",
       "      <td>0.084159</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>TriTraining (KNN)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.3</td>\n",
       "      <td>0.152725</td>\n",
       "      <td>0.092253</td>\n",
       "      <td>0.186408</td>\n",
       "      <td>0.063170</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>TriTraining (KNN)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0.146791</td>\n",
       "      <td>0.085206</td>\n",
       "      <td>0.182644</td>\n",
       "      <td>0.061573</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>TriTraining (KNN)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0.478078</td>\n",
       "      <td>0.061992</td>\n",
       "      <td>0.443366</td>\n",
       "      <td>0.035022</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>TriTraining (KNN)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.2</td>\n",
       "      <td>0.647748</td>\n",
       "      <td>0.070364</td>\n",
       "      <td>0.668955</td>\n",
       "      <td>0.020381</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>TriTraining (KNN)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.3</td>\n",
       "      <td>0.737913</td>\n",
       "      <td>0.080060</td>\n",
       "      <td>0.744788</td>\n",
       "      <td>0.015202</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>TriTraining (KNN)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0.773348</td>\n",
       "      <td>0.083499</td>\n",
       "      <td>0.800413</td>\n",
       "      <td>0.033052</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>TriTraining (NB)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0.090718</td>\n",
       "      <td>0.035779</td>\n",
       "      <td>0.044423</td>\n",
       "      <td>0.044664</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>TriTraining (NB)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.2</td>\n",
       "      <td>0.085693</td>\n",
       "      <td>0.053914</td>\n",
       "      <td>0.043761</td>\n",
       "      <td>0.053617</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>TriTraining (NB)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.3</td>\n",
       "      <td>0.090472</td>\n",
       "      <td>0.058631</td>\n",
       "      <td>0.057842</td>\n",
       "      <td>0.047577</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>TriTraining (NB)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0.088085</td>\n",
       "      <td>0.057578</td>\n",
       "      <td>0.082292</td>\n",
       "      <td>0.041538</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>TriTraining (NB)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0.401126</td>\n",
       "      <td>0.300181</td>\n",
       "      <td>0.406665</td>\n",
       "      <td>0.275658</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>TriTraining (NB)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.2</td>\n",
       "      <td>0.278153</td>\n",
       "      <td>0.210628</td>\n",
       "      <td>0.251790</td>\n",
       "      <td>0.211221</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>TriTraining (NB)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.3</td>\n",
       "      <td>0.166366</td>\n",
       "      <td>0.116880</td>\n",
       "      <td>0.196640</td>\n",
       "      <td>0.110576</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>TriTraining (NB)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0.155030</td>\n",
       "      <td>0.097225</td>\n",
       "      <td>0.180885</td>\n",
       "      <td>0.097111</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>TriTraining (CART)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0.167789</td>\n",
       "      <td>0.080768</td>\n",
       "      <td>0.181449</td>\n",
       "      <td>0.061071</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>TriTraining (CART)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.2</td>\n",
       "      <td>0.197977</td>\n",
       "      <td>0.063281</td>\n",
       "      <td>0.178936</td>\n",
       "      <td>0.060470</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>TriTraining (CART)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.3</td>\n",
       "      <td>0.192957</td>\n",
       "      <td>0.041478</td>\n",
       "      <td>0.197925</td>\n",
       "      <td>0.011228</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>TriTraining (CART)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0.169758</td>\n",
       "      <td>0.070042</td>\n",
       "      <td>0.198067</td>\n",
       "      <td>0.009981</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>TriTraining (CART)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0.715766</td>\n",
       "      <td>0.152987</td>\n",
       "      <td>0.769580</td>\n",
       "      <td>0.041982</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>TriTraining (CART)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.2</td>\n",
       "      <td>0.896321</td>\n",
       "      <td>0.074975</td>\n",
       "      <td>0.918764</td>\n",
       "      <td>0.020228</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>TriTraining (CART)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.3</td>\n",
       "      <td>0.901727</td>\n",
       "      <td>0.068320</td>\n",
       "      <td>0.916243</td>\n",
       "      <td>0.027497</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>TriTraining (CART)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0.907132</td>\n",
       "      <td>0.064047</td>\n",
       "      <td>0.946829</td>\n",
       "      <td>0.014985</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>Self-Training (KNN)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0.242754</td>\n",
       "      <td>0.084157</td>\n",
       "      <td>0.237461</td>\n",
       "      <td>0.007264</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>Self-Training (KNN)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.2</td>\n",
       "      <td>0.230311</td>\n",
       "      <td>0.083290</td>\n",
       "      <td>0.233099</td>\n",
       "      <td>0.011785</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>Self-Training (KNN)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.3</td>\n",
       "      <td>0.229596</td>\n",
       "      <td>0.073373</td>\n",
       "      <td>0.232699</td>\n",
       "      <td>0.008617</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>Self-Training (KNN)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0.227203</td>\n",
       "      <td>0.074212</td>\n",
       "      <td>0.226223</td>\n",
       "      <td>0.008845</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>Self-Training (KNN)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0.549399</td>\n",
       "      <td>0.072864</td>\n",
       "      <td>0.506809</td>\n",
       "      <td>0.052777</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>Self-Training (KNN)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.2</td>\n",
       "      <td>0.696772</td>\n",
       "      <td>0.045663</td>\n",
       "      <td>0.692879</td>\n",
       "      <td>0.026361</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>Self-Training (KNN)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.3</td>\n",
       "      <td>0.748574</td>\n",
       "      <td>0.053008</td>\n",
       "      <td>0.737414</td>\n",
       "      <td>0.025927</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>Self-Training (KNN)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0.781381</td>\n",
       "      <td>0.057703</td>\n",
       "      <td>0.785715</td>\n",
       "      <td>0.030130</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>Self-Training (NB)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0.075651</td>\n",
       "      <td>0.023860</td>\n",
       "      <td>0.065025</td>\n",
       "      <td>0.004766</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>Self-Training (NB)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.2</td>\n",
       "      <td>0.086187</td>\n",
       "      <td>0.033949</td>\n",
       "      <td>0.085060</td>\n",
       "      <td>0.010414</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>Self-Training (NB)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.3</td>\n",
       "      <td>0.091458</td>\n",
       "      <td>0.036784</td>\n",
       "      <td>0.084710</td>\n",
       "      <td>0.006705</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>Self-Training (NB)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0.091689</td>\n",
       "      <td>0.040028</td>\n",
       "      <td>0.083315</td>\n",
       "      <td>0.005651</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36</th>\n",
       "      <td>Self-Training (NB)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0.676877</td>\n",
       "      <td>0.250911</td>\n",
       "      <td>0.645830</td>\n",
       "      <td>0.256829</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37</th>\n",
       "      <td>Self-Training (NB)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.2</td>\n",
       "      <td>0.196396</td>\n",
       "      <td>0.069305</td>\n",
       "      <td>0.194363</td>\n",
       "      <td>0.012955</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>38</th>\n",
       "      <td>Self-Training (NB)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.3</td>\n",
       "      <td>0.196396</td>\n",
       "      <td>0.069305</td>\n",
       "      <td>0.215726</td>\n",
       "      <td>0.040900</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>39</th>\n",
       "      <td>Self-Training (NB)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0.196396</td>\n",
       "      <td>0.069305</td>\n",
       "      <td>0.223930</td>\n",
       "      <td>0.035261</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>40</th>\n",
       "      <td>Self-Training (CART)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0.208044</td>\n",
       "      <td>0.057041</td>\n",
       "      <td>0.210919</td>\n",
       "      <td>0.012029</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>41</th>\n",
       "      <td>Self-Training (CART)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.2</td>\n",
       "      <td>0.212577</td>\n",
       "      <td>0.058447</td>\n",
       "      <td>0.210920</td>\n",
       "      <td>0.010081</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42</th>\n",
       "      <td>Self-Training (CART)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.3</td>\n",
       "      <td>0.209482</td>\n",
       "      <td>0.045617</td>\n",
       "      <td>0.211227</td>\n",
       "      <td>0.010464</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>43</th>\n",
       "      <td>Self-Training (CART)</td>\n",
       "      <td>abalone</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0.212609</td>\n",
       "      <td>0.051943</td>\n",
       "      <td>0.210881</td>\n",
       "      <td>0.015029</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>44</th>\n",
       "      <td>Self-Training (CART)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0.778153</td>\n",
       "      <td>0.107618</td>\n",
       "      <td>0.790551</td>\n",
       "      <td>0.068355</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45</th>\n",
       "      <td>Self-Training (CART)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.2</td>\n",
       "      <td>0.879730</td>\n",
       "      <td>0.089891</td>\n",
       "      <td>0.911953</td>\n",
       "      <td>0.027151</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>46</th>\n",
       "      <td>Self-Training (CART)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.3</td>\n",
       "      <td>0.877102</td>\n",
       "      <td>0.071579</td>\n",
       "      <td>0.901929</td>\n",
       "      <td>0.026938</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>47</th>\n",
       "      <td>Self-Training (CART)</td>\n",
       "      <td>dermatology</td>\n",
       "      <td>0.4</td>\n",
       "      <td>0.909835</td>\n",
       "      <td>0.068335</td>\n",
       "      <td>0.935674</td>\n",
       "      <td>0.015822</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              classifier      dataset  labeling_rate  test_mean  test_std  \\\n",
       "0      TriTraining (KNN)      abalone            0.1   0.159610  0.093613   \n",
       "1      TriTraining (KNN)      abalone            0.2   0.185034  0.084115   \n",
       "2      TriTraining (KNN)      abalone            0.3   0.152725  0.092253   \n",
       "3      TriTraining (KNN)      abalone            0.4   0.146791  0.085206   \n",
       "4      TriTraining (KNN)  dermatology            0.1   0.478078  0.061992   \n",
       "5      TriTraining (KNN)  dermatology            0.2   0.647748  0.070364   \n",
       "6      TriTraining (KNN)  dermatology            0.3   0.737913  0.080060   \n",
       "7      TriTraining (KNN)  dermatology            0.4   0.773348  0.083499   \n",
       "8       TriTraining (NB)      abalone            0.1   0.090718  0.035779   \n",
       "9       TriTraining (NB)      abalone            0.2   0.085693  0.053914   \n",
       "10      TriTraining (NB)      abalone            0.3   0.090472  0.058631   \n",
       "11      TriTraining (NB)      abalone            0.4   0.088085  0.057578   \n",
       "12      TriTraining (NB)  dermatology            0.1   0.401126  0.300181   \n",
       "13      TriTraining (NB)  dermatology            0.2   0.278153  0.210628   \n",
       "14      TriTraining (NB)  dermatology            0.3   0.166366  0.116880   \n",
       "15      TriTraining (NB)  dermatology            0.4   0.155030  0.097225   \n",
       "16    TriTraining (CART)      abalone            0.1   0.167789  0.080768   \n",
       "17    TriTraining (CART)      abalone            0.2   0.197977  0.063281   \n",
       "18    TriTraining (CART)      abalone            0.3   0.192957  0.041478   \n",
       "19    TriTraining (CART)      abalone            0.4   0.169758  0.070042   \n",
       "20    TriTraining (CART)  dermatology            0.1   0.715766  0.152987   \n",
       "21    TriTraining (CART)  dermatology            0.2   0.896321  0.074975   \n",
       "22    TriTraining (CART)  dermatology            0.3   0.901727  0.068320   \n",
       "23    TriTraining (CART)  dermatology            0.4   0.907132  0.064047   \n",
       "24   Self-Training (KNN)      abalone            0.1   0.242754  0.084157   \n",
       "25   Self-Training (KNN)      abalone            0.2   0.230311  0.083290   \n",
       "26   Self-Training (KNN)      abalone            0.3   0.229596  0.073373   \n",
       "27   Self-Training (KNN)      abalone            0.4   0.227203  0.074212   \n",
       "28   Self-Training (KNN)  dermatology            0.1   0.549399  0.072864   \n",
       "29   Self-Training (KNN)  dermatology            0.2   0.696772  0.045663   \n",
       "30   Self-Training (KNN)  dermatology            0.3   0.748574  0.053008   \n",
       "31   Self-Training (KNN)  dermatology            0.4   0.781381  0.057703   \n",
       "32    Self-Training (NB)      abalone            0.1   0.075651  0.023860   \n",
       "33    Self-Training (NB)      abalone            0.2   0.086187  0.033949   \n",
       "34    Self-Training (NB)      abalone            0.3   0.091458  0.036784   \n",
       "35    Self-Training (NB)      abalone            0.4   0.091689  0.040028   \n",
       "36    Self-Training (NB)  dermatology            0.1   0.676877  0.250911   \n",
       "37    Self-Training (NB)  dermatology            0.2   0.196396  0.069305   \n",
       "38    Self-Training (NB)  dermatology            0.3   0.196396  0.069305   \n",
       "39    Self-Training (NB)  dermatology            0.4   0.196396  0.069305   \n",
       "40  Self-Training (CART)      abalone            0.1   0.208044  0.057041   \n",
       "41  Self-Training (CART)      abalone            0.2   0.212577  0.058447   \n",
       "42  Self-Training (CART)      abalone            0.3   0.209482  0.045617   \n",
       "43  Self-Training (CART)      abalone            0.4   0.212609  0.051943   \n",
       "44  Self-Training (CART)  dermatology            0.1   0.778153  0.107618   \n",
       "45  Self-Training (CART)  dermatology            0.2   0.879730  0.089891   \n",
       "46  Self-Training (CART)  dermatology            0.3   0.877102  0.071579   \n",
       "47  Self-Training (CART)  dermatology            0.4   0.909835  0.068335   \n",
       "\n",
       "    trans_mean  trans_std  \n",
       "0     0.186119   0.063314  \n",
       "1     0.167961   0.084159  \n",
       "2     0.186408   0.063170  \n",
       "3     0.182644   0.061573  \n",
       "4     0.443366   0.035022  \n",
       "5     0.668955   0.020381  \n",
       "6     0.744788   0.015202  \n",
       "7     0.800413   0.033052  \n",
       "8     0.044423   0.044664  \n",
       "9     0.043761   0.053617  \n",
       "10    0.057842   0.047577  \n",
       "11    0.082292   0.041538  \n",
       "12    0.406665   0.275658  \n",
       "13    0.251790   0.211221  \n",
       "14    0.196640   0.110576  \n",
       "15    0.180885   0.097111  \n",
       "16    0.181449   0.061071  \n",
       "17    0.178936   0.060470  \n",
       "18    0.197925   0.011228  \n",
       "19    0.198067   0.009981  \n",
       "20    0.769580   0.041982  \n",
       "21    0.918764   0.020228  \n",
       "22    0.916243   0.027497  \n",
       "23    0.946829   0.014985  \n",
       "24    0.237461   0.007264  \n",
       "25    0.233099   0.011785  \n",
       "26    0.232699   0.008617  \n",
       "27    0.226223   0.008845  \n",
       "28    0.506809   0.052777  \n",
       "29    0.692879   0.026361  \n",
       "30    0.737414   0.025927  \n",
       "31    0.785715   0.030130  \n",
       "32    0.065025   0.004766  \n",
       "33    0.085060   0.010414  \n",
       "34    0.084710   0.006705  \n",
       "35    0.083315   0.005651  \n",
       "36    0.645830   0.256829  \n",
       "37    0.194363   0.012955  \n",
       "38    0.215726   0.040900  \n",
       "39    0.223930   0.035261  \n",
       "40    0.210919   0.012029  \n",
       "41    0.210920   0.010081  \n",
       "42    0.211227   0.010464  \n",
       "43    0.210881   0.015029  \n",
       "44    0.790551   0.068355  \n",
       "45    0.911953   0.027151  \n",
       "46    0.901929   0.026938  \n",
       "47    0.935674   0.015822  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th colspan=\"4\" halign=\"left\">test_mean</th>\n",
       "      <th colspan=\"4\" halign=\"left\">test_std</th>\n",
       "      <th colspan=\"4\" halign=\"left\">trans_mean</th>\n",
       "      <th colspan=\"4\" halign=\"left\">trans_std</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>labeling_rate</th>\n",
       "      <th>0.1</th>\n",
       "      <th>0.2</th>\n",
       "      <th>0.3</th>\n",
       "      <th>0.4</th>\n",
       "      <th>0.1</th>\n",
       "      <th>0.2</th>\n",
       "      <th>0.3</th>\n",
       "      <th>0.4</th>\n",
       "      <th>0.1</th>\n",
       "      <th>0.2</th>\n",
       "      <th>0.3</th>\n",
       "      <th>0.4</th>\n",
       "      <th>0.1</th>\n",
       "      <th>0.2</th>\n",
       "      <th>0.3</th>\n",
       "      <th>0.4</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dataset</th>\n",
       "      <th>classifier</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"6\" valign=\"top\">abalone</th>\n",
       "      <th>Self-Training (CART)</th>\n",
       "      <td>0.208044</td>\n",
       "      <td>0.212577</td>\n",
       "      <td>0.209482</td>\n",
       "      <td>0.212609</td>\n",
       "      <td>0.057041</td>\n",
       "      <td>0.058447</td>\n",
       "      <td>0.045617</td>\n",
       "      <td>0.051943</td>\n",
       "      <td>0.210919</td>\n",
       "      <td>0.210920</td>\n",
       "      <td>0.211227</td>\n",
       "      <td>0.210881</td>\n",
       "      <td>0.012029</td>\n",
       "      <td>0.010081</td>\n",
       "      <td>0.010464</td>\n",
       "      <td>0.015029</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Self-Training (KNN)</th>\n",
       "      <td>0.242754</td>\n",
       "      <td>0.230311</td>\n",
       "      <td>0.229596</td>\n",
       "      <td>0.227203</td>\n",
       "      <td>0.084157</td>\n",
       "      <td>0.083290</td>\n",
       "      <td>0.073373</td>\n",
       "      <td>0.074212</td>\n",
       "      <td>0.237461</td>\n",
       "      <td>0.233099</td>\n",
       "      <td>0.232699</td>\n",
       "      <td>0.226223</td>\n",
       "      <td>0.007264</td>\n",
       "      <td>0.011785</td>\n",
       "      <td>0.008617</td>\n",
       "      <td>0.008845</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Self-Training (NB)</th>\n",
       "      <td>0.075651</td>\n",
       "      <td>0.086187</td>\n",
       "      <td>0.091458</td>\n",
       "      <td>0.091689</td>\n",
       "      <td>0.023860</td>\n",
       "      <td>0.033949</td>\n",
       "      <td>0.036784</td>\n",
       "      <td>0.040028</td>\n",
       "      <td>0.065025</td>\n",
       "      <td>0.085060</td>\n",
       "      <td>0.084710</td>\n",
       "      <td>0.083315</td>\n",
       "      <td>0.004766</td>\n",
       "      <td>0.010414</td>\n",
       "      <td>0.006705</td>\n",
       "      <td>0.005651</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TriTraining (CART)</th>\n",
       "      <td>0.167789</td>\n",
       "      <td>0.197977</td>\n",
       "      <td>0.192957</td>\n",
       "      <td>0.169758</td>\n",
       "      <td>0.080768</td>\n",
       "      <td>0.063281</td>\n",
       "      <td>0.041478</td>\n",
       "      <td>0.070042</td>\n",
       "      <td>0.181449</td>\n",
       "      <td>0.178936</td>\n",
       "      <td>0.197925</td>\n",
       "      <td>0.198067</td>\n",
       "      <td>0.061071</td>\n",
       "      <td>0.060470</td>\n",
       "      <td>0.011228</td>\n",
       "      <td>0.009981</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TriTraining (KNN)</th>\n",
       "      <td>0.159610</td>\n",
       "      <td>0.185034</td>\n",
       "      <td>0.152725</td>\n",
       "      <td>0.146791</td>\n",
       "      <td>0.093613</td>\n",
       "      <td>0.084115</td>\n",
       "      <td>0.092253</td>\n",
       "      <td>0.085206</td>\n",
       "      <td>0.186119</td>\n",
       "      <td>0.167961</td>\n",
       "      <td>0.186408</td>\n",
       "      <td>0.182644</td>\n",
       "      <td>0.063314</td>\n",
       "      <td>0.084159</td>\n",
       "      <td>0.063170</td>\n",
       "      <td>0.061573</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TriTraining (NB)</th>\n",
       "      <td>0.090718</td>\n",
       "      <td>0.085693</td>\n",
       "      <td>0.090472</td>\n",
       "      <td>0.088085</td>\n",
       "      <td>0.035779</td>\n",
       "      <td>0.053914</td>\n",
       "      <td>0.058631</td>\n",
       "      <td>0.057578</td>\n",
       "      <td>0.044423</td>\n",
       "      <td>0.043761</td>\n",
       "      <td>0.057842</td>\n",
       "      <td>0.082292</td>\n",
       "      <td>0.044664</td>\n",
       "      <td>0.053617</td>\n",
       "      <td>0.047577</td>\n",
       "      <td>0.041538</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"6\" valign=\"top\">dermatology</th>\n",
       "      <th>Self-Training (CART)</th>\n",
       "      <td>0.778153</td>\n",
       "      <td>0.879730</td>\n",
       "      <td>0.877102</td>\n",
       "      <td>0.909835</td>\n",
       "      <td>0.107618</td>\n",
       "      <td>0.089891</td>\n",
       "      <td>0.071579</td>\n",
       "      <td>0.068335</td>\n",
       "      <td>0.790551</td>\n",
       "      <td>0.911953</td>\n",
       "      <td>0.901929</td>\n",
       "      <td>0.935674</td>\n",
       "      <td>0.068355</td>\n",
       "      <td>0.027151</td>\n",
       "      <td>0.026938</td>\n",
       "      <td>0.015822</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Self-Training (KNN)</th>\n",
       "      <td>0.549399</td>\n",
       "      <td>0.696772</td>\n",
       "      <td>0.748574</td>\n",
       "      <td>0.781381</td>\n",
       "      <td>0.072864</td>\n",
       "      <td>0.045663</td>\n",
       "      <td>0.053008</td>\n",
       "      <td>0.057703</td>\n",
       "      <td>0.506809</td>\n",
       "      <td>0.692879</td>\n",
       "      <td>0.737414</td>\n",
       "      <td>0.785715</td>\n",
       "      <td>0.052777</td>\n",
       "      <td>0.026361</td>\n",
       "      <td>0.025927</td>\n",
       "      <td>0.030130</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Self-Training (NB)</th>\n",
       "      <td>0.676877</td>\n",
       "      <td>0.196396</td>\n",
       "      <td>0.196396</td>\n",
       "      <td>0.196396</td>\n",
       "      <td>0.250911</td>\n",
       "      <td>0.069305</td>\n",
       "      <td>0.069305</td>\n",
       "      <td>0.069305</td>\n",
       "      <td>0.645830</td>\n",
       "      <td>0.194363</td>\n",
       "      <td>0.215726</td>\n",
       "      <td>0.223930</td>\n",
       "      <td>0.256829</td>\n",
       "      <td>0.012955</td>\n",
       "      <td>0.040900</td>\n",
       "      <td>0.035261</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TriTraining (CART)</th>\n",
       "      <td>0.715766</td>\n",
       "      <td>0.896321</td>\n",
       "      <td>0.901727</td>\n",
       "      <td>0.907132</td>\n",
       "      <td>0.152987</td>\n",
       "      <td>0.074975</td>\n",
       "      <td>0.068320</td>\n",
       "      <td>0.064047</td>\n",
       "      <td>0.769580</td>\n",
       "      <td>0.918764</td>\n",
       "      <td>0.916243</td>\n",
       "      <td>0.946829</td>\n",
       "      <td>0.041982</td>\n",
       "      <td>0.020228</td>\n",
       "      <td>0.027497</td>\n",
       "      <td>0.014985</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TriTraining (KNN)</th>\n",
       "      <td>0.478078</td>\n",
       "      <td>0.647748</td>\n",
       "      <td>0.737913</td>\n",
       "      <td>0.773348</td>\n",
       "      <td>0.061992</td>\n",
       "      <td>0.070364</td>\n",
       "      <td>0.080060</td>\n",
       "      <td>0.083499</td>\n",
       "      <td>0.443366</td>\n",
       "      <td>0.668955</td>\n",
       "      <td>0.744788</td>\n",
       "      <td>0.800413</td>\n",
       "      <td>0.035022</td>\n",
       "      <td>0.020381</td>\n",
       "      <td>0.015202</td>\n",
       "      <td>0.033052</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TriTraining (NB)</th>\n",
       "      <td>0.401126</td>\n",
       "      <td>0.278153</td>\n",
       "      <td>0.166366</td>\n",
       "      <td>0.155030</td>\n",
       "      <td>0.300181</td>\n",
       "      <td>0.210628</td>\n",
       "      <td>0.116880</td>\n",
       "      <td>0.097225</td>\n",
       "      <td>0.406665</td>\n",
       "      <td>0.251790</td>\n",
       "      <td>0.196640</td>\n",
       "      <td>0.180885</td>\n",
       "      <td>0.275658</td>\n",
       "      <td>0.211221</td>\n",
       "      <td>0.110576</td>\n",
       "      <td>0.097111</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 test_mean                                \\\n",
       "labeling_rate                          0.1       0.2       0.3       0.4   \n",
       "dataset     classifier                                                     \n",
       "abalone     Self-Training (CART)  0.208044  0.212577  0.209482  0.212609   \n",
       "            Self-Training (KNN)   0.242754  0.230311  0.229596  0.227203   \n",
       "            Self-Training (NB)    0.075651  0.086187  0.091458  0.091689   \n",
       "            TriTraining (CART)    0.167789  0.197977  0.192957  0.169758   \n",
       "            TriTraining (KNN)     0.159610  0.185034  0.152725  0.146791   \n",
       "            TriTraining (NB)      0.090718  0.085693  0.090472  0.088085   \n",
       "dermatology Self-Training (CART)  0.778153  0.879730  0.877102  0.909835   \n",
       "            Self-Training (KNN)   0.549399  0.696772  0.748574  0.781381   \n",
       "            Self-Training (NB)    0.676877  0.196396  0.196396  0.196396   \n",
       "            TriTraining (CART)    0.715766  0.896321  0.901727  0.907132   \n",
       "            TriTraining (KNN)     0.478078  0.647748  0.737913  0.773348   \n",
       "            TriTraining (NB)      0.401126  0.278153  0.166366  0.155030   \n",
       "\n",
       "                                  test_std                                \\\n",
       "labeling_rate                          0.1       0.2       0.3       0.4   \n",
       "dataset     classifier                                                     \n",
       "abalone     Self-Training (CART)  0.057041  0.058447  0.045617  0.051943   \n",
       "            Self-Training (KNN)   0.084157  0.083290  0.073373  0.074212   \n",
       "            Self-Training (NB)    0.023860  0.033949  0.036784  0.040028   \n",
       "            TriTraining (CART)    0.080768  0.063281  0.041478  0.070042   \n",
       "            TriTraining (KNN)     0.093613  0.084115  0.092253  0.085206   \n",
       "            TriTraining (NB)      0.035779  0.053914  0.058631  0.057578   \n",
       "dermatology Self-Training (CART)  0.107618  0.089891  0.071579  0.068335   \n",
       "            Self-Training (KNN)   0.072864  0.045663  0.053008  0.057703   \n",
       "            Self-Training (NB)    0.250911  0.069305  0.069305  0.069305   \n",
       "            TriTraining (CART)    0.152987  0.074975  0.068320  0.064047   \n",
       "            TriTraining (KNN)     0.061992  0.070364  0.080060  0.083499   \n",
       "            TriTraining (NB)      0.300181  0.210628  0.116880  0.097225   \n",
       "\n",
       "                                 trans_mean                                \\\n",
       "labeling_rate                           0.1       0.2       0.3       0.4   \n",
       "dataset     classifier                                                      \n",
       "abalone     Self-Training (CART)   0.210919  0.210920  0.211227  0.210881   \n",
       "            Self-Training (KNN)    0.237461  0.233099  0.232699  0.226223   \n",
       "            Self-Training (NB)     0.065025  0.085060  0.084710  0.083315   \n",
       "            TriTraining (CART)     0.181449  0.178936  0.197925  0.198067   \n",
       "            TriTraining (KNN)      0.186119  0.167961  0.186408  0.182644   \n",
       "            TriTraining (NB)       0.044423  0.043761  0.057842  0.082292   \n",
       "dermatology Self-Training (CART)   0.790551  0.911953  0.901929  0.935674   \n",
       "            Self-Training (KNN)    0.506809  0.692879  0.737414  0.785715   \n",
       "            Self-Training (NB)     0.645830  0.194363  0.215726  0.223930   \n",
       "            TriTraining (CART)     0.769580  0.918764  0.916243  0.946829   \n",
       "            TriTraining (KNN)      0.443366  0.668955  0.744788  0.800413   \n",
       "            TriTraining (NB)       0.406665  0.251790  0.196640  0.180885   \n",
       "\n",
       "                                 trans_std                                \n",
       "labeling_rate                          0.1       0.2       0.3       0.4  \n",
       "dataset     classifier                                                    \n",
       "abalone     Self-Training (CART)  0.012029  0.010081  0.010464  0.015029  \n",
       "            Self-Training (KNN)   0.007264  0.011785  0.008617  0.008845  \n",
       "            Self-Training (NB)    0.004766  0.010414  0.006705  0.005651  \n",
       "            TriTraining (CART)    0.061071  0.060470  0.011228  0.009981  \n",
       "            TriTraining (KNN)     0.063314  0.084159  0.063170  0.061573  \n",
       "            TriTraining (NB)      0.044664  0.053617  0.047577  0.041538  \n",
       "dermatology Self-Training (CART)  0.068355  0.027151  0.026938  0.015822  \n",
       "            Self-Training (KNN)   0.052777  0.026361  0.025927  0.030130  \n",
       "            Self-Training (NB)    0.256829  0.012955  0.040900  0.035261  \n",
       "            TriTraining (CART)    0.041982  0.020228  0.027497  0.014985  \n",
       "            TriTraining (KNN)     0.035022  0.020381  0.015202  0.033052  \n",
       "            TriTraining (NB)      0.275658  0.211221  0.110576  0.097111  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.pivot_table(results, values=None, index=['dataset', 'classifier'], columns=['labeling_rate'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "------\n",
    "[1]: Isaac Triguero et. al \"Self-labeled techniques for semi-supervised learning:taxonomy, software and empirical study\" 2015\n",
    "\n",
    "[2]: Yarowsky D (1995) Unsupervised word sense disambiguation rivaling supervised methods. In: Proceedings\n",
    "of the 33rd annual meeting of the association for computational linguistics, pp 189–196\n",
    "\n",
    "[3]: Zhou ZH, Li M (2005) Tri-training: exploiting unlabeled data using three classifiers. IEEE Trans Knowl\n",
    "Data Eng 17:1529–1541"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Self-Training (CART) dermatology  Labeling ratio: 0.4\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaEAAAFoCAYAAAD+T0S+AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fdd35d21940>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Self-Training (CART) abalone  Labeling ratio: 0.4\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaMAAAFjCAYAAACQfWRAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fdd35ce4080>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TriTraining (NB) dermatology  Labeling ratio: 0.4\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaEAAAFoCAYAAAD+T0S+AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fdd70195d30>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Self-Training (NB) abalone  Labeling ratio: 0.4\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaMAAAFjCAYAAACQfWRAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fdd358f1390>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Self-Training (KNN) dermatology  Labeling ratio: 0.4\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaEAAAFhCAYAAADZQBV2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fdd357889e8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import seaborn as sns\n",
    "%matplotlib inline\n",
    "\n",
    "cnf_matrixes_to_show ={\n",
    "    (\"Self-Training (CART)\", \"dermatology\", 0.4),\n",
    "    (\"Self-Training (KNN)\", \"dermatology\", 0.4),\n",
    "    (\"Self-Training (CART)\", \"abalone\", 0.4),\n",
    "    (\"Self-Training (NB)\", \"abalone\", 0.4),\n",
    "    (\"TriTraining (NB)\", \"dermatology\", 0.4),\n",
    "}\n",
    "for classifier, dataset, ratio in cnf_matrixes_to_show:\n",
    "    matrix = cnf_matrixes[classifier][dataset][ratio]\n",
    "    print(classifier, dataset,\" Labeling ratio:\",ratio)\n",
    "    #print(matrix)\n",
    "    sns.heatmap(matrix,annot=True, square=True)\n",
    "    sns.plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "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": "3.5.2+"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}