fusion_two_organs.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from collections import defaultdict\n",
    "import numpy as np\n",
    "from matplotlib import pyplot as plt\n",
    "import matplotlib\n",
    "from collections import Counter\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.svm import SVC\n",
    "import random\n",
    "import matplotlib.patches as mpatches\n",
    "%matplotlib inline  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Late fusion approach between organ_1st and organ_2nd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "organ_1st = 'branch'\n",
    "organ_2nd = 'entire'\n",
    "top_k = 6 # top k scores that you want to know\n",
    "nb_class = 50\n",
    "flag_preprocess = False"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Preprocess for coding convenience"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "'''\n",
    "example:\n",
    "Change from:\n",
    "\n",
    "|organ 1              | organ 2               | organ 1 score     | organ 2 score    |\n",
    "|data/branch/123.jpg  | data/entire/256.jpg   | 0.6               | 0.4              |\n",
    "\n",
    "to:\n",
    "\n",
    "|organ 1              | organ 2               | organ 1 score     | organ 2 score    |\n",
    "|xxxxxxxxxxxxx        | xxxxxxxxxxxxx         | 0.6               | 0.4              |\n",
    "'''\n",
    "if (flag_preprocess):\n",
    "    f_1st = open('./fusion_data/' + 'single_organ_score/test_result_50_' +\n",
    "                 organ_1st + '_pretrained.txt', 'r')\n",
    "    f_2nd = open('./fusion_data/' + 'single_organ_score/test_result_50_' +\n",
    "                 organ_2nd + '_pretrained.txt', 'r')\n",
    "    fo_1st = open('./fusion_data/' + organ_1st + '_' + organ_2nd +\n",
    "                  '_50_species/test_result_50_' + organ_1st + '_pretrained.txt', 'w')\n",
    "    fo_2nd = open('./fusion_data/' + organ_1st + '_' + organ_2nd +\n",
    "                  '_50_species/test_result_50_' + organ_2nd + '_pretrained.txt', 'w')\n",
    "\n",
    "    count_line = 0\n",
    "    for line_1st in f_1st:\n",
    "        fo_1st.write(str(count_line/nb_class) + ' ' + line_1st.split(' ')[-3] +\n",
    "                      ' ' + line_1st.split(' ')[-2] +  ' ' + line_1st.split(' ')[-1])\n",
    "        count_line = count_line + 1\n",
    "    count_line = 0    \n",
    "    for line_2nd in f_2nd:\n",
    "        fo_2nd.write(str(count_line/nb_class) + ' ' + line_2nd.split(' ')[-3] +\n",
    "                      ' ' + line_2nd.split(' ')[-2] + ' ' + line_2nd.split(' ')[-1])\n",
    "        count_line = count_line + 1\n",
    "\n",
    "    f_1st.close()\n",
    "    f_2nd.close()\n",
    "    fo_1st.close()\n",
    "    fo_2nd.close()\n",
    "\n",
    "    f_1st = open('./fusion_data/' + 'single_organ_score/svm_result_50_' +\n",
    "                 organ_1st + '_pretrained.txt', 'r')\n",
    "    f_2nd = open('./fusion_data/' + 'single_organ_score/svm_result_50_' +\n",
    "                 organ_2nd + '_pretrained.txt', 'r')\n",
    "    fo_1st = open('./fusion_data/' + organ_1st + '_' + organ_2nd + '_50_species/svm_result_50_' +\n",
    "                  organ_1st + '_pretrained.txt', 'w')\n",
    "    fo_2nd = open('./fusion_data/' + organ_1st + '_' + organ_2nd + '_50_species/svm_result_50_' +\n",
    "                  organ_2nd + '_pretrained.txt', 'w')\n",
    "\n",
    "    count_line = 0   \n",
    "    for line_1st in f_1st:\n",
    "        fo_1st.write(str(count_line/nb_class) + ' ' + line_1st.split(' ')[-3] +\n",
    "                      ' ' + line_1st.split(' ')[-2] +  ' ' + line_1st.split(' ')[-1])\n",
    "        count_line = count_line + 1\n",
    "    count_line = 0    \n",
    "    for line_2nd in f_2nd:\n",
    "        fo_2nd.write(str(count_line/nb_class) + ' ' + line_2nd.split(' ')[-3] +\n",
    "                      ' ' + line_2nd.split(' ')[-2] + ' ' + line_2nd.split(' ')[-1])\n",
    "        count_line = count_line + 1\n",
    "\n",
    "    f_1st.close()\n",
    "    f_2nd.close()\n",
    "    fo_1st.close()\n",
    "    fo_2nd.close()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Using Sum, Product, Max rule"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "'''\n",
    "- sum rule: (score of first organ + score of second organ)\n",
    "- product rule: (score of first organ * score of second organ)\n",
    "- max rule: max(score of first organ , score of second organ)\n",
    "'''\n",
    "nb_class = 50\n",
    "\n",
    "f_1st = open('./fusion_data/' 'single_organ_score/test_result_50_' +\n",
    "             organ_1st + '_pretrained.txt', 'r')\n",
    "f_2nd = open('./fusion_data/' 'single_organ_score/test_result_50_' +\n",
    "             organ_2nd + '_pretrained.txt', 'r')\n",
    "\n",
    "score_1st = list()\n",
    "score_2nd = list()\n",
    "labels = list()\n",
    "\n",
    "for line in f_1st:\n",
    "    score_1st.append(float(line.split(' ')[-1]))\n",
    "    labels.append(int(line.split(' ')[-3]))\n",
    "    \n",
    "for line in f_2nd:\n",
    "    score_2nd.append(float(line.split(' ')[-1]))\n",
    "    \n",
    "st = list()\n",
    "nd = list()\n",
    "max_fu = list()\n",
    "sum_fu = list()\n",
    "prod_fu = list()\n",
    "\n",
    "max_fusion =  np.maximum(np.array(score_1st) , np.array(score_2nd))\n",
    "sum_fusion =  np.array(score_1st) + np.array(score_2nd)\n",
    "prod_fusion =  np.array(score_1st) * np.array(score_2nd)\n",
    "\n",
    "for k in range(1, top_k):\n",
    "    count_true_max = 0\n",
    "    count_true_sum = 0\n",
    "    count_true_prod = 0\n",
    "    count_true_1st = 0\n",
    "    count_true_2nd = 0\n",
    "    \n",
    "    for i in range(len(score_1st)/nb_class):\n",
    "        if labels[i * nb_class] in np.argsort(max_fusion[i * nb_class : (i + 1) * nb_class])[-k:]:\n",
    "            count_true_max = count_true_max + 1\n",
    "        \n",
    "        if labels[i * nb_class] in np.argsort(sum_fusion[i * nb_class : (i + 1) * nb_class])[-k:]:\n",
    "            count_true_sum = count_true_sum + 1\n",
    "            \n",
    "        if labels[i * nb_class] in np.argsort(prod_fusion[i * nb_class : (i + 1) * nb_class])[-k:]:\n",
    "            count_true_prod = count_true_prod + 1\n",
    "            \n",
    "        if labels[i * nb_class] in np.argsort(score_1st[i * nb_class : (i + 1) * nb_class])[-k:]:\n",
    "            count_true_1st = count_true_1st + 1\n",
    "\n",
    "        if labels[i * nb_class] in np.argsort(score_2nd[i * nb_class : (i + 1) * nb_class])[-k:]:\n",
    "            count_true_2nd = count_true_2nd + 1\n",
    "\n",
    "    max_fu.append(float(count_true_max) * nb_class/len(score_1st))\n",
    "    sum_fu.append(float(count_true_sum) * nb_class/len(score_1st))\n",
    "    prod_fu.append(float(count_true_prod) * nb_class/len(score_1st))\n",
    "    st.append(float(count_true_1st) * nb_class/len(score_1st))\n",
    "    nd.append(float(count_true_2nd) * nb_class/len(score_1st))\n",
    "\n",
    "f_1st.close()\n",
    "f_2nd.close()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Using  classification based SVM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "7350\n",
      "1500\n",
      "svm base: [0.538, 0.588, 0.612, 0.62, 0.626]\n"
     ]
    }
   ],
   "source": [
    "svm_fu = list()\n",
    "svm = SVC(kernel = 'poly', degree = 3, probability=True)\n",
    "\n",
    "ret_1st = defaultdict(list)\n",
    "ret_2nd = defaultdict(list)\n",
    "labels = list()\n",
    "points = list()\n",
    "svm_img_labels = dict()\n",
    "\n",
    "f_1st = open('./fusion_data/' + organ_1st + '_' + organ_2nd +\n",
    "             '_50_species/svm_result_50_' + organ_1st + '_pretrained.txt', 'r')\n",
    "for line in f_1st:\n",
    "    component = line.split(' ')\n",
    "    if int(component[2]) % nb_class == 0:\n",
    "        svm_img_labels[component[0].split('/')[-1]] = int(component[1])\n",
    "    ret_1st[component[0].split('/')[-1]].append(float(component[3]))\n",
    "f_1st.close()\n",
    "f_2nd = open('./fusion_data/' + organ_1st + '_' + organ_2nd +\n",
    "             '_50_species/svm_result_50_' + organ_2nd + '_pretrained.txt', 'r')\n",
    "for line in f_2nd:\n",
    "    component = line.split(' ')\n",
    "    ret_2nd[component[0].split('/')[-1]].append(float(component[3]))\n",
    "f_2nd.close()\n",
    "\n",
    "# Generate negative points\n",
    "for key in ret_1st.keys():\n",
    "    points.append([ret_1st[key][svm_img_labels[key]], ret_2nd[key][svm_img_labels[key]]])\n",
    "    labels.append(1)\n",
    "    for index in range(50):\n",
    "        if (index != svm_img_labels[key]):\n",
    "            points.append([ret_1st[key][index], ret_2nd[key][index]])\n",
    "            labels.append(0)\n",
    "\n",
    "svm.fit(points, labels)\n",
    "print(labels.count(0))\n",
    "print(labels.count(1))\n",
    "\n",
    "ret_1st_test = defaultdict(list)\n",
    "ret_2nd_test = defaultdict(list)\n",
    "test_img_labels = dict()\n",
    "\n",
    "f_1st_test = open('./fusion_data/' + organ_1st + '_' + organ_2nd +\n",
    "                  '_50_species/test_result_50_' + organ_1st + '_pretrained.txt', 'r')\n",
    "for line in f_1st_test:\n",
    "    component = line.split(' ')\n",
    "    if int(component[2]) % nb_class == 0:\n",
    "        test_img_labels[component[0].split('/')[-1]] = int(component[1])\n",
    "    ret_1st_test[component[0].split('/')[-1]].append(float(component[3]))\n",
    "f_1st_test.close()\n",
    "\n",
    "f_2nd_test = open('./fusion_data/' + organ_1st + '_' + organ_2nd +\n",
    "                  '_50_species/test_result_50_' + organ_2nd + '_pretrained.txt', 'r')\n",
    "for line in f_2nd_test:\n",
    "    component = line.split(' ')\n",
    "    ret_2nd_test[component[0].split('/')[-1]].append(float(component[3]))\n",
    "f_2nd_test.close()\n",
    "\n",
    "# Predict\n",
    "for k in range(1, top_k):\n",
    "    count_true = 0\n",
    "    tot_count = 0\n",
    "    for key in ret_1st_test.keys():\n",
    "        pos_list = list()\n",
    "        neg_list = list()\n",
    "        position = 0\n",
    "        for x, y in zip(ret_1st_test[key], ret_2nd_test[key]):\n",
    "            if svm.predict([[x, y]])[0] == 1:\n",
    "                pos_list.append([x, y, position])\n",
    "            else:\n",
    "                neg_list.append([x, y, position])\n",
    "            position = position + 1\n",
    "        \n",
    "        tot_count = tot_count + 1\n",
    "        if (pos_list):\n",
    "            pos_list.sort(key = lambda x: x[0] * x[1])\n",
    "            if test_img_labels[key] in np.asarray(pos_list[-k:])[:, 2]:\n",
    "                count_true = count_true + 1               \n",
    "        else:\n",
    "            neg_list.sort(key = lambda x:  x[0] * x[1])\n",
    "            if test_img_labels[key] in np.asarray(neg_list[-k:])[:, 2]:\n",
    "                count_true = count_true + 1\n",
    "    svm_fu.append(float(count_true)/(tot_count))\n",
    "\n",
    "print('svm base: ' + str(svm_fu))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 148,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYAAAAEZCAYAAACervI0AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nUEqpIKUJQCmlgpQmAKWUClL/DyvMBZi+RpUfAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1bad0dd8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot positive and negative points distribution\n",
    "fig, ax = plt.subplots()\n",
    "h = 0.001\n",
    "x_min, x_max = -0.1, 1.1\n",
    "y_min, y_max = -0.1, 1.1\n",
    "xx, yy = np.meshgrid(np.arange(x_min, x_max, h),\n",
    "                     np.arange(y_min, y_max, h))\n",
    "\n",
    "titles = 'SVC with polynomial (degree 4) kernel'\n",
    "\n",
    "Z = svm.predict(np.c_[xx.ravel(), yy.ravel()])\n",
    "# Put the result into a color plot\n",
    "Z = Z.reshape(xx.shape)\n",
    "plt.contourf(xx, yy, Z, cmap=plt.cm.Paired, alpha=0.8)\n",
    "\n",
    "\n",
    "colors = color= ['red' if l == 0 else 'green' for l in labels]\n",
    "plt.scatter(np.asarray(points)[:, 0], np.asarray(points)[:, 1], color = colors)\n",
    "\n",
    "classes = ['Positive','Negative']\n",
    "class_colours = ['green','red']\n",
    "recs = []\n",
    "for i in range(0,len(class_colours)):\n",
    "    recs.append(mpatches.Rectangle((0,0),1,1,fc=class_colours[i]))\n",
    "plt.legend(recs,classes,loc=4)\n",
    "plt.xlabel('branch score')\n",
    "plt.ylabel('leaf score')\n",
    "plt.title('Point distribution for all species')\n",
    "fig.savefig('all.eps', format='eps', dpi = 100)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####  Hybrid Model using Product rule and SVM classification probability"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "branch: [0.432, 0.532, 0.592, 0.638, 0.704]\n",
      "entire: [0.324, 0.442, 0.516, 0.578, 0.64]\n",
      "max rule: [0.516, 0.616, 0.682, 0.746, 0.768]\n",
      "sum rule: [0.522, 0.62, 0.694, 0.75, 0.776]\n",
      "prod rule: [0.58, 0.716, 0.77, 0.796, 0.836]\n",
      "svm base: [0.538, 0.588, 0.612, 0.62, 0.626]\n",
      "rh rule: [0.586, 0.716, 0.77, 0.806, 0.84]\n"
     ]
    }
   ],
   "source": [
    "rh_fu = list()\n",
    "svm = list()\n",
    "for i in range(nb_class):\n",
    "    svm.append(SVC(kernel = 'poly', degree = 3, probability=True))\n",
    "\n",
    "ret_1st = defaultdict(list)\n",
    "ret_2nd = defaultdict(list)\n",
    "labels = defaultdict(list)\n",
    "points = defaultdict(list)\n",
    "svm_img_labels = dict()\n",
    "\n",
    "f_1st = open('./fusion_data/' + organ_1st + '_' + organ_2nd + '_50_species/svm_result_50_' +\n",
    "             organ_1st + '_pretrained.txt', 'r')\n",
    "for line in f_1st:\n",
    "    component = line.split(' ')\n",
    "    if int(component[2]) % nb_class == 0:\n",
    "        svm_img_labels[component[0].split('/')[-1]] = int(component[1])\n",
    "    ret_1st[component[0].split('/')[-1]].append(float(component[3]))\n",
    "f_1st.close()\n",
    "f_2nd = open('./fusion_data/' + organ_1st + '_' + organ_2nd + '_50_species/svm_result_50_' +\n",
    "             organ_2nd + '_pretrained.txt', 'r')\n",
    "for line in f_2nd:\n",
    "    component = line.split(' ')\n",
    "    ret_2nd[component[0].split('/')[-1]].append(float(component[3]))\n",
    "f_2nd.close()\n",
    "\n",
    "rand_neg_point = random.randint(1, 10000) \n",
    "\n",
    "# Generate negative points\n",
    "for key in ret_1st.keys():\n",
    "    points[svm_img_labels[key]].append([ret_1st[key][svm_img_labels[key]], ret_2nd[key][svm_img_labels[key]]])\n",
    "    labels[svm_img_labels[key]].append(1)\n",
    "    for i in range(10):\n",
    "        rand_neg_point = random.randint(1, 10000) \n",
    "        index = rand_neg_point % nb_class\n",
    "        if (index != svm_img_labels[key]):\n",
    "            if labels[index].count(0) > 10 * labels[index].count(1):\n",
    "                continue\n",
    "            points[index].append([ret_1st[key][index], ret_2nd[key][index]])\n",
    "            labels[index].append(0)\n",
    "\n",
    "for i in range(nb_class):\n",
    "    svm[i].fit(points[i], labels[i])\n",
    "\n",
    "ret_1st_test = defaultdict(list)\n",
    "ret_2nd_test = defaultdict(list)\n",
    "test_img_labels = dict()\n",
    "\n",
    "f_1st_test = open('./fusion_data/' + organ_1st + '_' + organ_2nd + '_50_species/test_result_50_' +\n",
    "                  organ_1st + '_pretrained.txt', 'r')\n",
    "for line in f_1st_test:\n",
    "    component = line.split(' ')\n",
    "    if int(component[2]) % nb_class == 0:\n",
    "        test_img_labels[component[0].split('/')[-1]] = int(component[1])\n",
    "    ret_1st_test[component[0].split('/')[-1]].append(float(component[3]))\n",
    "f_1st_test.close()\n",
    "\n",
    "f_2nd_test = open('./fusion_data/' + organ_1st + '_' + organ_2nd + '_50_species/test_result_50_' +\n",
    "                  organ_2nd + '_pretrained.txt', 'r')\n",
    "for line in f_2nd_test:\n",
    "    component = line.split(' ')\n",
    "    ret_2nd_test[component[0].split('/')[-1]].append(float(component[3]))\n",
    "f_2nd_test.close()\n",
    "\n",
    "# Predict\n",
    "for k in range(1, top_k):\n",
    "    count_true = 0\n",
    "    tot_count = 0\n",
    "    for key in ret_1st_test.keys():\n",
    "        pos_list = list()\n",
    "        neg_list = list()\n",
    "        position = 0\n",
    "        for x, y in zip(ret_1st_test[key], ret_2nd_test[key]):\n",
    "            if (svm[position].predict_proba([[x, y]])[0][1] > 0):\n",
    "                pos_list.append([x, y, svm[position].predict_proba([[x, y]])[0][1], position])\n",
    "            else:\n",
    "                neg_list.append([x, y, svm[position].predict_proba([[x, y]])[0][1], position])\n",
    "            position = position + 1\n",
    "        \n",
    "        tot_count = tot_count + 1\n",
    "        if (pos_list):\n",
    "            pos_list.sort(key = lambda x: x[0] * x[1] * x[2])\n",
    "            if test_img_labels[key] in np.asarray(pos_list[-k:])[:, 3]:\n",
    "                count_true = count_true + 1               \n",
    "        else:\n",
    "            print('Negative point')\n",
    "            neg_list.sort(key = lambda x: (x[0] * x[1]))\n",
    "            if (neg_list[-1][3] == test_img_labels[key]):\n",
    "                count_true = count_true + 1\n",
    "                print('Bing go!')\n",
    "    rh_fu.append(float(count_true)/(tot_count))\n",
    "\n",
    "print(organ_1st + ': ' + str(st))\n",
    "print(organ_2nd + ': ' + str(nd))\n",
    "print('max rule: ' + str(max_fu))\n",
    "print('sum rule: ' + str(sum_fu))\n",
    "print('prod rule: ' + str(prod_fu))\n",
    "print('svm base: ' + str(svm_fu))\n",
    "print('rh rule: ' + str(rh_fu))\n",
    "\n",
    "output = open('./fusion_data/' + organ_1st + '_' + organ_2nd + '_top_' + str(top_k) + '_scores.txt', 'w')\n",
    "output.write(organ_1st + ': ' + str(st) + '\\n')\n",
    "output.write(organ_2nd + ': ' + str(nd) + '\\n')\n",
    "output.write('max rule: ' + str(max_fu) + '\\n')\n",
    "output.write('sum rule: ' + str(sum_fu) + '\\n')\n",
    "output.write('prod rule: ' + str(prod_fu) + '\\n')\n",
    "output.write('svm based: ' + str(svm_fu) + '\\n')\n",
    "output.write('rh rule: ' + str(rh_fu))\n",
    "output.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYAAAAEZCAYAAACervI0AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngIhIglICEBFJUEoAIiIJ6v8BvUmCogPbe5MAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0xb9c7e48>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot positve and negative points distribution for each species\n",
    "fig, ax = plt.subplots()\n",
    "id = 8\n",
    "h = 0.001\n",
    "x_min, x_max = -0.1, 1.1\n",
    "y_min, y_max = -0.1, 1.1\n",
    "xx, yy = np.meshgrid(np.arange(x_min, x_max, h),\n",
    "                     np.arange(y_min, y_max, h))\n",
    "\n",
    "Z = svm[id].predict(np.c_[xx.ravel(), yy.ravel()])\n",
    "# Put the result into a color plot\n",
    "Z = Z.reshape(xx.shape)\n",
    "plt.contourf(xx, yy, Z, cmap=plt.cm.Paired, alpha=0.8)\n",
    "\n",
    "\n",
    "colors = color= ['red' if l == 0 else 'green' for l in labels[id]]\n",
    "plt.scatter(np.asarray(points[id])[:, 0], np.asarray(points[id])[:, 1], color = colors)\n",
    "\n",
    "classes = ['Positive','Negative']\n",
    "class_colours = ['green','red']\n",
    "recs = []\n",
    "for i in range(0,len(class_colours)):\n",
    "    recs.append(mpatches.Rectangle((0,0),1,1,fc=class_colours[i]))\n",
    "plt.legend(recs,classes,loc=4)\n",
    "plt.xlabel('branch score')\n",
    "plt.ylabel('leaf score')\n",
    "plt.title('Point distribution for species %d' %id)\n",
    "fig.savefig('one.eps', format='eps', dpi=1000)\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}