NewF1.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import sys\n",
    "sep = os.sep\n",
    "import numpy as np\n",
    "# from testarch.unet import runs as ru\n",
    "from testarch.miniunet import runs as ru\n",
    "import utils.auto_split as asp"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "from PIL import Image as IMG\n",
    "from utils.img_utils import Image\n",
    "from sklearn.metrics import precision_recall_fscore_support\n",
    "from scipy.ndimage.morphology import distance_transform_edt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "runs_unet = [\n",
    "    ru.DRIVE_WEIGHTED, ru.DRIVE_1_1, ru.DRIVE_1_100_1,\n",
    "    ru.STARE_WEIGHTED, ru.STARE_1_1, ru.STARE_1_100_1,\n",
    "    ru.WIDE_WEIGHTED, ru.WIDE_1_1, ru.WIDE_1_100_1,\n",
    "    ru.VEVIO_MOSAICS_WEIGHTED, ru.VEVIO_MOSAICS_1_1, ru.VEVIO_MOSAICS_1_100_1,\n",
    "    ru.VEVIO_FRAMES_WEIGHTED, ru.VEVIO_FRAMES_1_1, ru.VEVIO_FRAMES_1_100_1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "log_file = 'F1'+sep+'DRIVE_MINIUNET.csv'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "logs\\DRIVE\\MINI_UNET_WEIGHTED\n",
      "### SPLIT FOUND:  data\\DRIVE\\splits\\DRIVE.json Loaded\n",
      "logs\\DRIVE\\MINI_UNET_1__1\n",
      "### SPLIT FOUND:  data\\DRIVE\\splits\\DRIVE.json Loaded\n",
      "logs\\DRIVE\\MINI_UNET_1_100_1\n",
      "### SPLIT FOUND:  data\\DRIVE\\splits\\DRIVE.json Loaded\n",
      "logs\\STARE\\MINI_UNET_WEIGHTED\n",
      "### SPLIT FOUND:  data\\STARE\\splits\\STARE_0.json Loaded\n",
      "### SPLIT FOUND:  data\\STARE\\splits\\STARE_1.json Loaded\n",
      "### SPLIT FOUND:  data\\STARE\\splits\\STARE_2.json Loaded\n",
      "### SPLIT FOUND:  data\\STARE\\splits\\STARE_3.json Loaded\n",
      "### SPLIT FOUND:  data\\STARE\\splits\\STARE_4.json Loaded\n",
      "logs\\STARE\\MINI_UNET_1__1\n",
      "### SPLIT FOUND:  data\\STARE\\splits\\STARE_0.json Loaded\n",
      "### SPLIT FOUND:  data\\STARE\\splits\\STARE_1.json Loaded\n",
      "### SPLIT FOUND:  data\\STARE\\splits\\STARE_2.json Loaded\n",
      "### SPLIT FOUND:  data\\STARE\\splits\\STARE_3.json Loaded\n",
      "### SPLIT FOUND:  data\\STARE\\splits\\STARE_4.json Loaded\n",
      "logs\\STARE\\MINI_UNET_1_100_1\n",
      "### SPLIT FOUND:  data\\STARE\\splits\\STARE_0.json Loaded\n",
      "### SPLIT FOUND:  data\\STARE\\splits\\STARE_1.json Loaded\n",
      "### SPLIT FOUND:  data\\STARE\\splits\\STARE_2.json Loaded\n",
      "### SPLIT FOUND:  data\\STARE\\splits\\STARE_3.json Loaded\n",
      "### SPLIT FOUND:  data\\STARE\\splits\\STARE_4.json Loaded\n",
      "logs\\AV_WIDE\\MINI_UNET_WEIGHTED\n",
      "### SPLIT FOUND:  data\\AV-WIDE\\splits\\AV-WIDE_0.json Loaded\n",
      "### SPLIT FOUND:  data\\AV-WIDE\\splits\\AV-WIDE_1.json Loaded\n",
      "### SPLIT FOUND:  data\\AV-WIDE\\splits\\AV-WIDE_2.json Loaded\n",
      "### SPLIT FOUND:  data\\AV-WIDE\\splits\\AV-WIDE_3.json Loaded\n",
      "### SPLIT FOUND:  data\\AV-WIDE\\splits\\AV-WIDE_4.json Loaded\n",
      "### SPLIT FOUND:  data\\AV-WIDE\\splits\\AV-WIDE_5.json Loaded\n",
      "logs\\AV_WIDE\\MINI_UNET_1__1\n",
      "### SPLIT FOUND:  data\\AV-WIDE\\splits\\AV-WIDE_0.json Loaded\n",
      "### SPLIT FOUND:  data\\AV-WIDE\\splits\\AV-WIDE_1.json Loaded\n",
      "### SPLIT FOUND:  data\\AV-WIDE\\splits\\AV-WIDE_2.json Loaded\n",
      "### SPLIT FOUND:  data\\AV-WIDE\\splits\\AV-WIDE_3.json Loaded\n",
      "### SPLIT FOUND:  data\\AV-WIDE\\splits\\AV-WIDE_4.json Loaded\n",
      "### SPLIT FOUND:  data\\AV-WIDE\\splits\\AV-WIDE_5.json Loaded\n",
      "logs\\AV_WIDE\\MINI_UNET_1_100_1\n",
      "### SPLIT FOUND:  data\\AV-WIDE\\splits\\AV-WIDE_0.json Loaded\n",
      "### SPLIT FOUND:  data\\AV-WIDE\\splits\\AV-WIDE_1.json Loaded\n",
      "### SPLIT FOUND:  data\\AV-WIDE\\splits\\AV-WIDE_2.json Loaded\n",
      "### SPLIT FOUND:  data\\AV-WIDE\\splits\\AV-WIDE_3.json Loaded\n",
      "### SPLIT FOUND:  data\\AV-WIDE\\splits\\AV-WIDE_4.json Loaded\n",
      "### SPLIT FOUND:  data\\AV-WIDE\\splits\\AV-WIDE_5.json Loaded\n",
      "logs\\VEVIO_MOSAICS\\MINI_UNET_WEIGHTED\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_mosaics\\VEVIO_0.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_mosaics\\VEVIO_1.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_mosaics\\VEVIO_2.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_mosaics\\VEVIO_3.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_mosaics\\VEVIO_4.json Loaded\n",
      "logs\\VEVIO_MOSAICS\\MINI_UNET_1__1\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_mosaics\\VEVIO_0.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_mosaics\\VEVIO_1.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_mosaics\\VEVIO_2.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_mosaics\\VEVIO_3.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_mosaics\\VEVIO_4.json Loaded\n",
      "logs\\VEVIO_MOSAICS\\MINI_UNET_1_100_1\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_mosaics\\VEVIO_0.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_mosaics\\VEVIO_1.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_mosaics\\VEVIO_2.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_mosaics\\VEVIO_3.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_mosaics\\VEVIO_4.json Loaded\n",
      "logs\\VEVIO_FRAMES\\MINI_UNET_WEIGHTED\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_frames\\VEVIO_0.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_frames\\VEVIO_1.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_frames\\VEVIO_2.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_frames\\VEVIO_3.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_frames\\VEVIO_4.json Loaded\n",
      "logs\\VEVIO_FRAMES\\MINI_UNET_1__1\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_frames\\VEVIO_0.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_frames\\VEVIO_1.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_frames\\VEVIO_2.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_frames\\VEVIO_3.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_frames\\VEVIO_4.json Loaded\n",
      "logs\\VEVIO_FRAMES\\MINI_UNET_1_100_1\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_frames\\VEVIO_0.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_frames\\VEVIO_1.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_frames\\VEVIO_2.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_frames\\VEVIO_3.json Loaded\n",
      "### SPLIT FOUND:  data\\VEVIO\\splits_frames\\VEVIO_4.json Loaded\n"
     ]
    }
   ],
   "source": [
    "with open(log_file, 'w+') as log:\n",
    "    log.write('DATASET,EXPERIMENT,PREC_AVG,RECALL_AVG,F1')\n",
    "    for d in runs_unet:\n",
    "        print(d['Dirs']['logs'])\n",
    "        dset = d['Dirs']['logs'].split(os.sep)[1]\n",
    "        exp = d['Dirs']['logs'].split(os.sep)[-1]\n",
    "        results = {'p':[],'r':[], 'f1':[]}\n",
    "\n",
    "        test_files = []\n",
    "        for split_file in os.listdir(d['Dirs']['splits_json']):\n",
    "            test_files += asp.load_split_json(d['Dirs']['splits_json']+sep+split_file)['test']\n",
    "\n",
    "        for file in test_files:\n",
    "            image = Image()\n",
    "            image.load_file(d['Dirs']['image'], file)\n",
    "            image.load_ground_truth(d['Dirs']['truth'], d['Funcs']['truth_getter'])\n",
    "\n",
    "            if d['Dirs'].get('mask'):\n",
    "                image.load_mask(d['Dirs']['mask'], d['Funcs']['mask_getter'])\n",
    "            \n",
    "            try:\n",
    "                pred_file =  'pred_' + file.split('.')[0] + '.png'\n",
    "                image.pred = np.array(IMG.open(d['Dirs']['logs'] + sep + pred_file))\n",
    "            except:\n",
    "                pred_file =  file.split('.')[0] + '.png'\n",
    "                image.pred = np.array(IMG.open(d['Dirs']['logs'] + sep + pred_file))\n",
    "\n",
    "            image.pred[image.pred==255] = 1\n",
    "\n",
    "            gt = image.ground_truth.copy()\n",
    "            gt[gt==255] = 1\n",
    "            edt_gt = distance_transform_edt(gt)\n",
    "\n",
    "            edt = edt_gt.copy()\n",
    "            edt[edt>2] = 0\n",
    "            \n",
    "            ### Supress thick vessels when computing F1 Score\n",
    "            gt_masked = gt.copy()\n",
    "            gt_masked[edt==0] = 0\n",
    "            \n",
    "            pred_masked = image.pred.copy()\n",
    "            pred_masked[edt==0] = 0\n",
    "\n",
    "            p, r, f1, _ = precision_recall_fscore_support(gt.flatten(), image.pred.flatten(), average='binary')\n",
    "            results['p'].append(p)\n",
    "            results['r'].append(r)\n",
    "            results['f1'].append(f1)\n",
    "        p_avg = np.array(results['p']).mean()\n",
    "        r_avg = np.array(results['r']).mean()\n",
    "        f1 =  (2 * p_avg * r_avg) / (p_avg + r_avg)\n",
    "        line = '{},{},{},{},{}'.format(dset, exp, round(p_avg, 5), round(r_avg, 5), round(f1, 5))\n",
    "        log.write('\\n'+line)\n",
    "        log.flush()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "import json"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_splits(files, k=0, json_file='SPLIT', shuffle_files=True):\n",
    "    import json\n",
    "    from random import shuffle\n",
    "    from itertools import chain\n",
    "    \n",
    "    json_file = json_file.split('.')[0] + '.json'\n",
    "    if shuffle_files:\n",
    "        shuffle(files)\n",
    "        \n",
    "    ix_splits =  np.array_split(np.arange(len(files)), k)\n",
    "    for i in range(len(ix_splits)):\n",
    "        \n",
    "        test_ix = ix_splits[i].tolist()\n",
    "        val_ix = ix_splits[(i+1)%len(ix_splits)].tolist()\n",
    "        train_ix = [ix for ix in np.arange(len(files)) if ix not in test_ix + val_ix]\n",
    "        \n",
    "        splits = {}\n",
    "        splits['train'] = [files[ix] for ix in train_ix]\n",
    "        splits['validation'] = [files[ix] for ix in val_ix]\n",
    "        splits['test'] = [files[ix] for ix in test_ix]\n",
    "        \n",
    "        print('Valid:', set(list(chain(*splits.values())))- set(files)==set([]))\n",
    "        \n",
    "        f = open(str(i) + json_file, \"w\")\n",
    "        f.write(json.dumps(splits))\n",
    "        f.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Valid: True\n",
      "Valid: True\n",
      "Valid: True\n",
      "Valid: True\n",
      "Valid: True\n"
     ]
    }
   ],
   "source": [
    "create_splits(os.listdir(ru.DRIVE_1_100_1['Dirs']['image']), k=5, json_file='CHASEDB')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [],
   "source": [
    "def dice_loss(outputs=None, target=None, beta=1, weights=None,):\n",
    "    \"\"\"\n",
    "    :param weights: element-wise weights\n",
    "    :param outputs:\n",
    "    :param target:\n",
    "    :param beta: More beta, better precision. 1 is neutral\n",
    "    :return:\n",
    "    \"\"\"\n",
    "    smooth = 1.\n",
    "    if weights is not None:\n",
    "        weights = weights.contiguous().float().view(-1)\n",
    "    else:\n",
    "        weights = 1\n",
    "\n",
    "    iflat = outputs.contiguous().float().view(-1)\n",
    "    tflat = target.contiguous().float().view(-1)\n",
    "    intersection = (iflat * tflat * weights).sum()\n",
    "\n",
    "    f = (((1 + beta ** 2) * intersection) + smooth) / (((beta ** 2 * iflat.sum()) + tflat.sum()) + smooth)\n",
    "    return 1 - f\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 190,
   "metadata": {},
   "outputs": [],
   "source": [
    "a = torch.Tensor([1, 1, 0, 1, 0])\n",
    "b = torch.Tensor([1, 0, 1, 1, 1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 204,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor(0.3750)"
      ]
     },
     "execution_count": 204,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dice_loss(a, b, 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 192,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_true = b.clone().int().view(1, -1).squeeze()\n",
    "y_pred = a.clone().int().view(1, -1).squeeze()\n",
    "\n",
    "y_true[y_true == 255] = 1\n",
    "y_pred[y_pred == 255] = 1\n",
    "\n",
    "y_true = y_true * 2\n",
    "y_cases = y_true + y_pred\n",
    "tp = torch.sum(y_cases == 3).item()\n",
    "fp = torch.sum(y_cases == 1).item()\n",
    "tn = torch.sum(y_cases == 0).item()\n",
    "fn = torch.sum(y_cases == 2).item()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 209,
   "metadata": {},
   "outputs": [],
   "source": [
    "beta  = 0\n",
    "try:\n",
    "    p = tp / (tp + fp)\n",
    "except ZeroDivisionError:\n",
    "    p = 0\n",
    "try:\n",
    "    r = tp / (tp + fn)\n",
    "except ZeroDivisionError:\n",
    "    r = 0\n",
    "try:\n",
    "    f = (1 + beta ** 2) * p * r / (((beta ** 2) * p) + r)\n",
    "except ZeroDivisionError:\n",
    "    f = 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 212,
   "metadata": {},
   "outputs": [],
   "source": [
    "import random\n",
    "# Lets define our Beta Function to generate s for any particular state. We don't care for the normalizing constant here.\n",
    "def beta_s(w,a,b):\n",
    "    return w**(a-1)*(1-w)**(b-1)\n",
    "\n",
    "# This Function returns True if the coin with probability P of heads comes heads when flipped.\n",
    "def random_coin(p):\n",
    "    unif = random.uniform(0,1)\n",
    "    if unif>=p:\n",
    "        return False\n",
    "    else:\n",
    "        return True\n",
    "\n",
    "# This Function runs the MCMC chain for Beta Distribution.\n",
    "def beta_mcmc(N_hops,a,b):\n",
    "    states = []\n",
    "    cur = random.uniform(0,1)\n",
    "    for i in range(0,N_hops):\n",
    "        states.append(cur)\n",
    "        next = random.uniform(0,1)\n",
    "        ap = min(beta_s(next,a,b)/beta_s(cur,a,b),1) # Calculate the acceptance probability\n",
    "        if random_coin(ap):\n",
    "            cur = next\n",
    "    return states[-1000:] # Returns the last 100 states of the chain"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 214,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\python37\\lib\\site-packages\\ipykernel_launcher.py:12: RuntimeWarning: divide by zero encountered in double_scalars\n",
      "  if sys.path[0] == '':\n",
      "c:\\python37\\lib\\site-packages\\ipykernel_launcher.py:13: RuntimeWarning: divide by zero encountered in double_scalars\n",
      "  del sys.path[0]\n",
      "c:\\python37\\lib\\site-packages\\ipykernel_launcher.py:25: MatplotlibDeprecationWarning: \n",
      "The 'normed' kwarg was deprecated in Matplotlib 2.1 and will be removed in 3.1. Use 'density' instead.\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1224x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1224x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1224x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1224x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import numpy as np\n",
    "import pylab as pl\n",
    "import scipy.special as ss\n",
    "%matplotlib inline\n",
    "pl.rcParams['figure.figsize'] = (17.0, 4.0)\n",
    "\n",
    "# Actual Beta PDF.\n",
    "def beta(a, b, i):\n",
    "    e1 = ss.gamma(a + b)\n",
    "    e2 = ss.gamma(a)\n",
    "    e3 = ss.gamma(b)\n",
    "    e4 = i ** (a - 1)\n",
    "    e5 = (1 - i) ** (b - 1)\n",
    "    return (e1/(e2*e3)) * e4 * e5\n",
    "\n",
    "# Create a function to plot Actual Beta PDF with the Beta Sampled from MCMC Chain.\n",
    "def plot_beta(a, b):\n",
    "    Ly = []\n",
    "    Lx = []\n",
    "    i_list = np.mgrid[0:1:100j]\n",
    "    for i in i_list:\n",
    "        Lx.append(i)\n",
    "        Ly.append(beta(a, b, i))\n",
    "    pl.plot(Lx, Ly, label=\"Real Distribution: a=\"+str(a)+\", b=\"+str(b))\n",
    "    pl.hist(beta_mcmc(100000,a,b),normed=True,bins =25, histtype='step',label=\"Simulated_MCMC: a=\"+str(a)+\", b=\"+str(b))\n",
    "    pl.legend()\n",
    "    pl.show()\n",
    "    \n",
    "plot_beta(0.1, 0.1)\n",
    "plot_beta(1, 1)\n",
    "plot_beta(2, 3)\n",
    "plot_beta(0.5, 0.5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 197,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.6666666666666666, 0.5, 0.5048543689320388)"
      ]
     },
     "execution_count": 197,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "p, r, f"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 246,
   "metadata": {},
   "outputs": [
    {
     "ename": "SyntaxError",
     "evalue": "invalid syntax (<ipython-input-246-b4e289aa5450>, line 1)",
     "output_type": "error",
     "traceback": [
      "\u001b[1;36m  File \u001b[1;32m\"<ipython-input-246-b4e289aa5450>\"\u001b[1;36m, line \u001b[1;32m1\u001b[0m\n\u001b[1;33m    a = 1==1 5 else 10\u001b[0m\n\u001b[1;37m             ^\u001b[0m\n\u001b[1;31mSyntaxError\u001b[0m\u001b[1;31m:\u001b[0m invalid syntax\n"
     ]
    }
   ],
   "source": [
    "a = 1==1 5 else 10"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}