modification pipeline 14-15 - spindle detection and training.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['64chlocs.elp', 'label_extraction.npy', 'label_extraction.pkl', 'pineline 13.csv', 'pipe line 10', 'pipe line 10.csv', 'pipeline 13.pickle', 's5d2_final.edf', 's5d2_final_annotations.txt', 's6n2_final.edf', 's6n2_final_annotations.txt', 'Sleep Stage Scoring Criteria.docx', 'spindle training.p', 'suj10_d1final.edf', 'suj10_d1final_annotations.txt', 'suj10_d2final.edf', 'suj10_d2final_annotations.txt', 'suj13_l2nap_day2 edited.edf', 'suj13_l2nap_day2 edited1.edf', 'suj13_l2nap_day2 edited1_annotations.txt', 'suj13_l2nap_day2 edited_annotations.txt', 'suj13_l2nap_day2 edited_C3.txt', 'suj13_l2nap_day2 edited_C4.txt', 'suj13_l2nap_day2 edited_F3.txt', 'suj13_l2nap_day2 edited_F4.txt', 'suj13_l2nap_day2 edited_O1.txt', 'suj13_l2nap_day2 edited_O2.txt', 'suj5_d1final.edf', 'suj5_d1final_annotations.txt', 'suj6_d1final.edf', 'suj6_d1final_annotations.txt', 'suj8_d1final.edf', 'suj8_d1final_annotations.txt', 'suj8_d2final.edf', 'suj8_d2final_annotations.txt', 'suj9_d1final.edf', 'suj9_d1final_annotations.txt', 'suj9_d2final.edf', 'suj9_d2final_annotations.txt', 'testing-montage-2.mtg', 'Training Data Frame']\n"
     ]
    }
   ],
   "source": [
    "import eegPinelineDesign\n",
    "import numpy as np\n",
    "import random\n",
    "import mne\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "plt.style.use('ggplot') \n",
    "import os\n",
    "import pandas as pd\n",
    "import re\n",
    "import json\n",
    "import scipy\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.decomposition import PCA,FastICA\n",
    "from sklearn.pipeline import Pipeline,make_pipeline\n",
    "from sklearn.grid_search import GridSearchCV\n",
    "from sklearn.cross_validation import cross_val_score\n",
    "from scipy.fftpack import fft,ifft\n",
    "import math\n",
    "from sklearn.metrics import classification_report,accuracy_score,confusion_matrix\n",
    "from scipy.signal import spectrogram,find_peaks_cwt\n",
    "from mne.preprocessing.ica import ICA\n",
    "from sklearn.metrics import precision_recall_curve\n",
    "from sklearn.metrics import average_precision_score\n",
    "from sklearn.cross_validation import train_test_split,ShuffleSplit\n",
    "from sklearn.preprocessing import label_binarize,scale\n",
    "from sklearn.multiclass import OneVsRestClassifier\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.preprocessing import label_binarize,StandardScaler\n",
    "from nitime import algorithms as alg\n",
    "eegPinelineDesign.change_file_directory('C:/Users/ning/Downloads/training set')\n",
    "EDFfiles, Annotationfiles = eegPinelineDesign.split_type_of_files()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "resolution = 1;spindle_training_data=[];non_spindle_training_data=[];spindle_labels=[];non_spindle_labels=[]\n",
    "\n",
    "# where two intervals overlap\n",
    "def getOverlap(a,b):\n",
    "    return max(0,min(a[1],b[1]) - max(a[0],b[0]))\n",
    "\n",
    "result = {}\n",
    "for num, EDFfileName in enumerate(EDFfiles):\n",
    "    if EDFfileName == 'suj13_l2nap_day2 edited.edf' or EDFfileName =='suj13_l2nap_day2 edited1.edf':\n",
    "        pass\n",
    "    else:\n",
    "        file_to_read,fileName=eegPinelineDesign.pick_sample_file(EDFfiles,n=num)\n",
    "        # channel list for artifact(eyemovement or chewing) removal - ROC and/or LOC\n",
    "        channelList = ['F3','F4','C3','C4','O1','O2','ROC','LOC']\n",
    "        raw = eegPinelineDesign.load_data(file_to_read,channelList,None, None)# No pass\n",
    "        annotation_to_read = [x for x in Annotationfiles if fileName in x]\n",
    "        file = pd.read_csv(annotation_to_read[0])\n",
    "        labelFind = re.compile('spindle',eegPinelineDesign.re.IGNORECASE)\n",
    "        spindles=[]# take existed annotations\n",
    "        for row in file.iterrows():\n",
    "            currentEvent = row[1][-1]\n",
    "            if labelFind.search(currentEvent):\n",
    "                spindles.append(row[1][0])# time of marker\n",
    "        print('finish spindle annotation')\n",
    "        # threshold test to detect possible spindle segments\n",
    "        channelList=['F3','F4','C3','C4','O1','O2']\n",
    "        windowsize = 5;ratio_index={};muscle_index={};sample_ind={}\n",
    "        print('start threshold')\n",
    "        for ii,names in enumerate(channelList):\n",
    "            #initialize\n",
    "            endPoint=windowsize;startPoint=endPoint-windowsize\n",
    "            distance_to_end = raw.last_samp/1000 - endPoint\n",
    "            ratio_index[names] =[];muscle_index[names]=[];sample_ind[names]=[];\n",
    "\n",
    "            # start the while loop\n",
    "            print('start the while loop')\n",
    "            while distance_to_end > windowsize:\n",
    "                start, stop = raw.time_as_index([startPoint,endPoint])\n",
    "                tempSegment, time = raw[ii,start:stop]\n",
    "\n",
    "\n",
    "\n",
    "                filter_alpha = mne.filter.band_pass_filter(tempSegment[0,:],1000,8,12.5) #band pass alpha\n",
    "                RMS_alpha = np.sqrt(sum(filter_alpha**2)/len(filter_alpha))\n",
    "\n",
    "                filter_spindle = mne.filter.band_pass_filter(tempSegment[0,:],1000,11.5,16.5)# band pass 11.5 - 16\n",
    "                RMS_spindle = np.sqrt(sum(filter_spindle**2)/len(filter_spindle))\n",
    "\n",
    "                filter_muscle = mne.filter.band_pass_filter(tempSegment[0,:],1000,30,40.5) # band pass 30 - 40, muscle movement\n",
    "                RMS_muscle = np.sqrt(sum(filter_muscle**2)/len(filter_muscle))\n",
    "\n",
    "                point=[0,0]\n",
    "                if RMS_alpha/RMS_spindle < 1.2:# very subjective value\n",
    "                    point[0] += 2\n",
    "                    ratio_index[names].append(2)\n",
    "                else:\n",
    "                    point[0] += -1\n",
    "                    ratio_index[names].append(-1)\n",
    "\n",
    "                if RMS_muscle > 5*10e-4:\n",
    "                    point[1] += -1\n",
    "                    muscle_index[names].append(-1)\n",
    "                else:\n",
    "                    point[1] += 1\n",
    "                    muscle_index[names].append(1)\n",
    "\n",
    "                sample_ind[names].append(sum(point))\n",
    "\n",
    "                endPoint += resolution\n",
    "                startPoint=endPoint - windowsize\n",
    "                distance_to_end = raw.last_samp/1000 - endPoint\n",
    "            sample_ind[names].append(0)\n",
    "        general_result = pd.DataFrame(sample_ind)\n",
    "        print('finish the while loop')\n",
    "        mark_pairs={};indicator={};print('start mark pairs')\n",
    "        for ii,names in enumerate(channelList):\n",
    "            mark_pairs[names]=[];indicator[names]=[]\n",
    "            for pos,items in enumerate(general_result[names]):\n",
    "                if pos == len(general_result[names])-1:\n",
    "                    pass\n",
    "                else:\n",
    "\n",
    "                    indicator[names].append(general_result[names][pos+1]-general_result[names][pos])\n",
    "\n",
    "                    if indicator[names][pos] == -3:\n",
    "                        endPoint = pos*resolution\n",
    "                        initial_position=pos;position_idx=pos\n",
    "                        try:\n",
    "                            while indicator[names][initial_position] != 3:\n",
    "                                position_idx -= 1 # backward search\n",
    "                                initial_position = position_idx\n",
    "                            startPoint = position_idx * resolution\n",
    "\n",
    "                            if endPoint - startPoint >= 1:\n",
    "                                mark_pairs[names].append((startPoint,endPoint))\n",
    "\n",
    "                        except:\n",
    "                            pass\n",
    "\n",
    "        # start to put in spindle event and non events\n",
    "        print('start putting in spindle instances')\n",
    "        cnt = 0\n",
    "        for instance in spindles:\n",
    "            tempData=[]\n",
    "            for ii, names in enumerate(channelList):\n",
    "                startPoint=instance-2.5;endPoint=instance+2.5\n",
    "                start,stop=raw.time_as_index([startPoint,endPoint])\n",
    "                tempSegment,timeSpan = raw[ii,start:stop]\n",
    "                centerxval=timeSpan[np.argmax(abs(tempSegment[0,:]))]\n",
    "                startPoint=centerxval-1.5;endPoint=centerxval+1.5\n",
    "                start,stop=raw.time_as_index([startPoint,endPoint])\n",
    "                tempSegment,time = raw[ii,start:stop]\n",
    "                tempData = np.append(tempData,tempSegment)\n",
    "            spindle_training_data.append(tempData)\n",
    "            spindle_labels.append(1)\n",
    "            cnt +=1\n",
    "        print('start puting in non spindle instances')\n",
    "        used_time_windows=[]\n",
    "        for ii in range(cnt):\n",
    "            Condition_operator=[False,False,False,False,False,False,False,False]\n",
    "\n",
    "            while sum(np.array(Condition_operator).astype(int)) < 7:\n",
    "                instance = np.random.choice(np.arange(raw.last_samp/1000),1)\n",
    "                startPoint = instance - 1.5; endPoint = instance + 1.5\n",
    "\n",
    "                for idx, names in enumerate(channelList):\n",
    "                    Condition_operator[idx]=any(getOverlap([instance -1,instance + 1],[lower,upper]) !=0 for (lower,upper) in mark_pairs[names])\n",
    "\n",
    "                Condition_operator[6]=all(getOverlap([startPoint,endPoint],[instance-1.5,instance+1.5])==0 for instance in spindles)\n",
    "                Condition_operator[7]=all(getOverlap([startPoint,endPoint],[lower,upper])!=0 for (lower,upper) in used_time_windows)\n",
    "                print(Condition_operator)\n",
    "            used_time_windows.append([startPoint,endPoint])\n",
    "            tempData=[]\n",
    "            for idxx,names in enumerate(channelList):\n",
    "                startPoint=instance-1.5;endPoint=instance+1.5\n",
    "                start,stop=raw.time_as_index([startPoint,endPoint])\n",
    "                tempSegment,time = raw[idxx,start:stop]\n",
    "                tempData = np.append(tempData,tempSegment)\n",
    "            non_spindle_training_data.append(tempData)\n",
    "            non_spindle_labels.append(0)\n",
    "            print(ii)\n",
    "        print(EDFfileName,len(spindle_training_data),len(non_spindle_training_data))\n",
    "        result[EDFfileName]={'spindle training':spindle_training_data,'spindle label':spindle_labels,\n",
    "                            'non spindle training':non_spindle_training_data,'non spindle label':non_spindle_labels}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "import pickle\n",
    "with open('spindle training.p','rb') as handle:\n",
    "    pickle.dump(result,handle)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['64chlocs.elp', 'label_extraction.npy', 'label_extraction.pkl', 'pineline 13.csv', 'pipe line 10', 'pipe line 10.csv', 'pipeline 13.pickle', 's5d2_final.edf', 's5d2_final_annotations.txt', 's6n2_final.edf', 's6n2_final_annotations.txt', 'Sleep Stage Scoring Criteria.docx', 'spindle training.p', 'suj10_d1final.edf', 'suj10_d1final_annotations.txt', 'suj10_d2final.edf', 'suj10_d2final_annotations.txt', 'suj13_l2nap_day2 edited.edf', 'suj13_l2nap_day2 edited1.edf', 'suj13_l2nap_day2 edited1_annotations.txt', 'suj13_l2nap_day2 edited_annotations.txt', 'suj13_l2nap_day2 edited_C3.txt', 'suj13_l2nap_day2 edited_C4.txt', 'suj13_l2nap_day2 edited_F3.txt', 'suj13_l2nap_day2 edited_F4.txt', 'suj13_l2nap_day2 edited_O1.txt', 'suj13_l2nap_day2 edited_O2.txt', 'suj5_d1final.edf', 'suj5_d1final_annotations.txt', 'suj6_d1final.edf', 'suj6_d1final_annotations.txt', 'suj8_d1final.edf', 'suj8_d1final_annotations.txt', 'suj8_d2final.edf', 'suj8_d2final_annotations.txt', 'suj9_d1final.edf', 'suj9_d1final_annotations.txt', 'suj9_d2final.edf', 'suj9_d2final_annotations.txt', 'testing-montage-2.mtg', 'Training Data Frame']\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Anaconda3\\lib\\site-packages\\sklearn\\lda.py:4: DeprecationWarning: lda.LDA has been moved to discriminant_analysis.LinearDiscriminantAnalysis in 0.17 and will be removed in 0.19\n",
      "  \"in 0.17 and will be removed in 0.19\", DeprecationWarning)\n"
     ]
    }
   ],
   "source": [
    "import eegPinelineDesign\n",
    "import numpy as np\n",
    "import random\n",
    "import mne\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "plt.style.use('ggplot') \n",
    "import os\n",
    "import pandas as pd\n",
    "import re\n",
    "import json\n",
    "import scipy\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.lda import LDA\n",
    "from sklearn.decomposition import PCA,FastICA\n",
    "from sklearn.pipeline import Pipeline,make_pipeline\n",
    "from sklearn.grid_search import GridSearchCV\n",
    "from sklearn.cross_validation import cross_val_score\n",
    "from scipy.fftpack import fft,ifft\n",
    "import math\n",
    "from sklearn.metrics import classification_report,accuracy_score,confusion_matrix\n",
    "from scipy.signal import spectrogram,find_peaks_cwt\n",
    "from mne.preprocessing.ica import ICA\n",
    "from sklearn.metrics import precision_recall_curve\n",
    "from sklearn.metrics import average_precision_score\n",
    "from sklearn.cross_validation import train_test_split,ShuffleSplit\n",
    "from sklearn.preprocessing import label_binarize,scale\n",
    "from sklearn.multiclass import OneVsRestClassifier\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.preprocessing import label_binarize,StandardScaler\n",
    "from nitime import algorithms as alg\n",
    "eegPinelineDesign.change_file_directory('C:/Users/ning/Downloads/training set')\n",
    "EDFfiles, Annotationfiles = eegPinelineDesign.split_type_of_files()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import pickle\n",
    "with open('spindle training.p','rb') as handle:\n",
    "    result=pickle.load(handle)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['suj8_d1final.edf', 'suj5_d1final.edf', 's5d2_final.edf', 's6n2_final.edf', 'suj9_d2final.edf', 'suj10_d1final.edf', 'suj9_d1final.edf', 'suj6_d1final.edf', 'suj8_d2final.edf', 'suj10_d2final.edf'])"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "suj8_d1final.edf 1363\n",
      "suj5_d1final.edf 1363\n",
      "s5d2_final.edf 1363\n",
      "s6n2_final.edf 1363\n",
      "suj9_d2final.edf 1363\n",
      "suj10_d1final.edf 1363\n",
      "suj9_d1final.edf 1363\n",
      "suj6_d1final.edf 1363\n",
      "suj8_d2final.edf 1363\n",
      "suj10_d2final.edf 1363\n"
     ]
    }
   ],
   "source": [
    "for items in result.items():\n",
    "    print(items[0],len(items[1]['spindle training']))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1363\n",
      "1363\n",
      "1363\n",
      "1363\n"
     ]
    }
   ],
   "source": [
    "spindle_training_data = result['suj9_d1final.edf']['spindle training'];print(len(spindle_training_data))\n",
    "non_spindle_training_data = result['suj9_d1final.edf']['non spindle training'];print(len(non_spindle_training_data))\n",
    "spindle_label=result['suj9_d1final.edf']['spindle label'];print(len(spindle_label))\n",
    "non_spindle_label=result['suj9_d1final.edf']['non spindle label'];print(len(non_spindle_label))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1364 1364\n",
      "1365 1365\n",
      "1366 1366\n",
      "1367 1367\n",
      "1368 1368\n",
      "1369 1369\n",
      "1370 1370\n",
      "1371 1371\n",
      "1372 1372\n",
      "1373 1373\n",
      "1374 1374\n",
      "1375 1375\n",
      "1376 1376\n",
      "1377 1377\n",
      "1378 1378\n",
      "1379 1379\n",
      "1380 1380\n",
      "1381 1381\n",
      "1382 1382\n",
      "1383 1383\n",
      "1384 1384\n",
      "1385 1385\n",
      "1386 1386\n",
      "1387 1387\n",
      "1388 1388\n",
      "1389 1389\n",
      "1390 1390\n",
      "1391 1391\n",
      "1392 1392\n",
      "1393 1393\n",
      "1394 1394\n",
      "1395 1395\n",
      "1396 1396\n",
      "1397 1397\n",
      "1398 1398\n",
      "1399 1399\n",
      "1400 1400\n",
      "1401 1401\n",
      "1402 1402\n",
      "1403 1403\n",
      "1404 1404\n",
      "1405 1405\n",
      "1406 1406\n",
      "1407 1407\n",
      "1408 1408\n",
      "1409 1409\n",
      "1410 1410\n",
      "1411 1411\n",
      "1412 1412\n",
      "1413 1413\n",
      "1414 1414\n",
      "1415 1415\n",
      "1416 1416\n",
      "1417 1417\n",
      "1418 1418\n",
      "1419 1419\n",
      "1420 1420\n",
      "1421 1421\n",
      "1422 1422\n",
      "1423 1423\n",
      "1424 1424\n",
      "1425 1425\n",
      "1426 1426\n",
      "1427 1427\n",
      "1428 1428\n",
      "1429 1429\n",
      "1430 1430\n",
      "1431 1431\n",
      "1432 1432\n",
      "1433 1433\n",
      "1434 1434\n",
      "1435 1435\n",
      "1436 1436\n",
      "1437 1437\n",
      "1438 1438\n",
      "1439 1439\n",
      "1440 1440\n",
      "1441 1441\n",
      "1442 1442\n",
      "1443 1443\n",
      "1444 1444\n",
      "1445 1445\n",
      "1446 1446\n",
      "1447 1447\n",
      "1448 1448\n",
      "1449 1449\n",
      "1450 1450\n",
      "1451 1451\n",
      "1452 1452\n",
      "1453 1453\n",
      "1454 1454\n",
      "1455 1455\n",
      "1456 1456\n",
      "1457 1457\n",
      "1458 1458\n",
      "1459 1459\n",
      "1460 1460\n",
      "1461 1461\n",
      "1462 1462\n",
      "1463 1463\n",
      "1464 1464\n",
      "1465 1465\n",
      "1466 1466\n",
      "1467 1467\n",
      "1468 1468\n",
      "1469 1469\n",
      "1470 1470\n",
      "1471 1471\n",
      "1472 1472\n",
      "1473 1473\n",
      "1474 1474\n",
      "1475 1475\n",
      "1476 1476\n",
      "1477 1477\n",
      "1478 1478\n",
      "1479 1479\n",
      "1480 1480\n",
      "1481 1481\n",
      "1482 1482\n",
      "1483 1483\n",
      "1484 1484\n",
      "1485 1485\n",
      "1486 1486\n",
      "1487 1487\n",
      "1488 1488\n",
      "1489 1489\n",
      "1490 1490\n",
      "1491 1491\n",
      "1492 1492\n",
      "1493 1493\n",
      "1494 1494\n",
      "1495 1495\n",
      "1496 1496\n",
      "1497 1497\n",
      "1498 1498\n",
      "1499 1499\n",
      "1500 1500\n",
      "1501 1501\n",
      "1502 1502\n",
      "1503 1503\n",
      "1504 1504\n",
      "1505 1505\n",
      "1506 1506\n",
      "1507 1507\n",
      "1508 1508\n",
      "1509 1509\n",
      "1510 1510\n",
      "1511 1511\n",
      "1512 1512\n",
      "1513 1513\n",
      "1514 1514\n",
      "1515 1515\n",
      "1516 1516\n",
      "1517 1517\n",
      "1518 1518\n",
      "1519 1519\n",
      "1520 1520\n",
      "1521 1521\n",
      "1522 1522\n",
      "1523 1523\n",
      "1524 1524\n",
      "1525 1525\n",
      "1526 1526\n",
      "1527 1527\n",
      "1528 1528\n",
      "1529 1529\n",
      "1530 1530\n",
      "1531 1531\n",
      "1532 1532\n",
      "1533 1533\n",
      "1534 1534\n",
      "1535 1535\n",
      "1536 1536\n",
      "1537 1537\n",
      "1538 1538\n",
      "1539 1539\n",
      "1540 1540\n",
      "1541 1541\n",
      "1542 1542\n",
      "1543 1543\n",
      "1544 1544\n",
      "1545 1545\n",
      "1546 1546\n",
      "1547 1547\n",
      "1548 1548\n",
      "1549 1549\n",
      "1550 1550\n",
      "1551 1551\n",
      "1552 1552\n",
      "1553 1553\n",
      "1554 1554\n",
      "1555 1555\n",
      "1556 1556\n",
      "1557 1557\n",
      "1558 1558\n",
      "1559 1559\n",
      "1560 1560\n",
      "1561 1561\n",
      "1562 1562\n",
      "1563 1563\n",
      "1564 1564\n",
      "1565 1565\n",
      "1566 1566\n",
      "1567 1567\n",
      "1568 1568\n",
      "1569 1569\n",
      "1570 1570\n",
      "1571 1571\n",
      "1572 1572\n",
      "1573 1573\n",
      "1574 1574\n",
      "1575 1575\n",
      "1576 1576\n",
      "1577 1577\n",
      "1578 1578\n",
      "1579 1579\n",
      "1580 1580\n",
      "1581 1581\n",
      "1582 1582\n",
      "1583 1583\n",
      "1584 1584\n",
      "1585 1585\n",
      "1586 1586\n",
      "1587 1587\n",
      "1588 1588\n",
      "1589 1589\n",
      "1590 1590\n",
      "1591 1591\n",
      "1592 1592\n",
      "1593 1593\n",
      "1594 1594\n",
      "1595 1595\n",
      "1596 1596\n",
      "1597 1597\n",
      "1598 1598\n",
      "1599 1599\n",
      "1600 1600\n",
      "1601 1601\n",
      "1602 1602\n",
      "1603 1603\n",
      "1604 1604\n",
      "1605 1605\n",
      "1606 1606\n",
      "1607 1607\n",
      "1608 1608\n",
      "1609 1609\n",
      "1610 1610\n",
      "1611 1611\n",
      "1612 1612\n",
      "1613 1613\n",
      "1614 1614\n",
      "1615 1615\n",
      "1616 1616\n",
      "1617 1617\n",
      "1618 1618\n",
      "1619 1619\n",
      "1620 1620\n",
      "1621 1621\n",
      "1622 1622\n",
      "1623 1623\n",
      "1624 1624\n",
      "1625 1625\n",
      "1626 1626\n",
      "1627 1627\n",
      "1628 1628\n",
      "1629 1629\n",
      "1630 1630\n",
      "1631 1631\n",
      "1632 1632\n",
      "1633 1633\n",
      "1634 1634\n",
      "1635 1635\n",
      "1636 1636\n",
      "1637 1637\n",
      "1638 1638\n",
      "1639 1639\n",
      "1640 1640\n",
      "1641 1641\n",
      "1642 1642\n",
      "1643 1643\n",
      "1644 1644\n",
      "1645 1645\n",
      "1646 1646\n",
      "1647 1647\n",
      "1648 1648\n",
      "1649 1649\n",
      "1650 1650\n",
      "1651 1651\n",
      "1652 1652\n",
      "1653 1653\n",
      "1654 1654\n",
      "1655 1655\n",
      "1656 1656\n",
      "1657 1657\n",
      "1658 1658\n",
      "1659 1659\n",
      "1660 1660\n",
      "1661 1661\n",
      "1662 1662\n",
      "1663 1663\n",
      "1664 1664\n",
      "1665 1665\n",
      "1666 1666\n",
      "1667 1667\n",
      "1668 1668\n",
      "1669 1669\n",
      "1670 1670\n",
      "1671 1671\n",
      "1672 1672\n",
      "1673 1673\n",
      "1674 1674\n",
      "1675 1675\n",
      "1676 1676\n",
      "1677 1677\n",
      "1678 1678\n",
      "1679 1679\n",
      "1680 1680\n",
      "1681 1681\n",
      "1682 1682\n",
      "1683 1683\n",
      "1684 1684\n",
      "1685 1685\n",
      "1686 1686\n",
      "1687 1687\n",
      "1688 1688\n",
      "1689 1689\n",
      "1690 1690\n",
      "1691 1691\n",
      "1692 1692\n",
      "1693 1693\n",
      "1694 1694\n",
      "1695 1695\n",
      "1696 1696\n",
      "1697 1697\n",
      "1698 1698\n",
      "1699 1699\n",
      "1700 1700\n",
      "1701 1701\n",
      "1702 1702\n",
      "1703 1703\n",
      "1704 1704\n",
      "1705 1705\n",
      "1706 1706\n",
      "1707 1707\n",
      "1708 1708\n",
      "1709 1709\n",
      "1710 1710\n",
      "1711 1711\n",
      "1712 1712\n",
      "1713 1713\n",
      "1714 1714\n",
      "1715 1715\n",
      "1716 1716\n",
      "1717 1717\n",
      "1718 1718\n",
      "1719 1719\n",
      "1720 1720\n",
      "1721 1721\n",
      "1722 1722\n",
      "1723 1723\n",
      "1724 1724\n",
      "1725 1725\n",
      "1726 1726\n",
      "1727 1727\n",
      "1728 1728\n",
      "1729 1729\n",
      "1730 1730\n",
      "1731 1731\n",
      "1732 1732\n",
      "1733 1733\n",
      "1734 1734\n",
      "1735 1735\n",
      "1736 1736\n",
      "1737 1737\n",
      "1738 1738\n",
      "1739 1739\n",
      "1740 1740\n",
      "1741 1741\n",
      "1742 1742\n",
      "1743 1743\n",
      "1744 1744\n",
      "1745 1745\n",
      "1746 1746\n",
      "1747 1747\n",
      "1748 1748\n",
      "1749 1749\n",
      "1750 1750\n",
      "1751 1751\n",
      "1752 1752\n",
      "1753 1753\n",
      "1754 1754\n",
      "1755 1755\n",
      "1756 1756\n",
      "1757 1757\n",
      "1758 1758\n",
      "1759 1759\n",
      "1760 1760\n",
      "1761 1761\n",
      "1762 1762\n",
      "1763 1763\n",
      "1764 1764\n",
      "1765 1765\n",
      "1766 1766\n",
      "1767 1767\n",
      "1768 1768\n",
      "1769 1769\n",
      "1770 1770\n",
      "1771 1771\n",
      "1772 1772\n",
      "1773 1773\n",
      "1774 1774\n",
      "1775 1775\n",
      "1776 1776\n",
      "1777 1777\n",
      "1778 1778\n",
      "1779 1779\n",
      "1780 1780\n",
      "1781 1781\n",
      "1782 1782\n",
      "1783 1783\n",
      "1784 1784\n",
      "1785 1785\n",
      "1786 1786\n",
      "1787 1787\n",
      "1788 1788\n",
      "1789 1789\n",
      "1790 1790\n",
      "1791 1791\n",
      "1792 1792\n",
      "1793 1793\n",
      "1794 1794\n",
      "1795 1795\n",
      "1796 1796\n",
      "1797 1797\n",
      "1798 1798\n",
      "1799 1799\n",
      "1800 1800\n",
      "1801 1801\n",
      "1802 1802\n",
      "1803 1803\n",
      "1804 1804\n",
      "1805 1805\n",
      "1806 1806\n",
      "1807 1807\n",
      "1808 1808\n",
      "1809 1809\n",
      "1810 1810\n",
      "1811 1811\n",
      "1812 1812\n",
      "1813 1813\n",
      "1814 1814\n",
      "1815 1815\n",
      "1816 1816\n",
      "1817 1817\n",
      "1818 1818\n",
      "1819 1819\n",
      "1820 1820\n",
      "1821 1821\n",
      "1822 1822\n",
      "1823 1823\n",
      "1824 1824\n",
      "1825 1825\n",
      "1826 1826\n",
      "1827 1827\n",
      "1828 1828\n",
      "1829 1829\n",
      "1830 1830\n",
      "1831 1831\n",
      "1832 1832\n",
      "1833 1833\n",
      "1834 1834\n",
      "1835 1835\n",
      "1836 1836\n",
      "1837 1837\n",
      "1838 1838\n",
      "1839 1839\n",
      "1840 1840\n",
      "1841 1841\n",
      "1842 1842\n",
      "1843 1843\n",
      "1844 1844\n",
      "1845 1845\n",
      "1846 1846\n",
      "1847 1847\n",
      "1848 1848\n",
      "1849 1849\n",
      "1850 1850\n",
      "1851 1851\n",
      "1852 1852\n",
      "1853 1853\n",
      "1854 1854\n",
      "1855 1855\n",
      "1856 1856\n",
      "1857 1857\n",
      "1858 1858\n",
      "1859 1859\n",
      "1860 1860\n",
      "1861 1861\n",
      "1862 1862\n",
      "1863 1863\n",
      "1864 1864\n",
      "1865 1865\n",
      "1866 1866\n",
      "1867 1867\n",
      "1868 1868\n",
      "1869 1869\n",
      "1870 1870\n",
      "1871 1871\n",
      "1872 1872\n",
      "1873 1873\n",
      "1874 1874\n",
      "1875 1875\n",
      "1876 1876\n",
      "1877 1877\n",
      "1878 1878\n",
      "1879 1879\n",
      "1880 1880\n",
      "1881 1881\n",
      "1882 1882\n",
      "1883 1883\n",
      "1884 1884\n",
      "1885 1885\n",
      "1886 1886\n",
      "1887 1887\n",
      "1888 1888\n",
      "1889 1889\n",
      "1890 1890\n",
      "1891 1891\n",
      "1892 1892\n",
      "1893 1893\n",
      "1894 1894\n",
      "1895 1895\n",
      "1896 1896\n",
      "1897 1897\n",
      "1898 1898\n",
      "1899 1899\n",
      "1900 1900\n",
      "1901 1901\n",
      "1902 1902\n",
      "1903 1903\n",
      "1904 1904\n",
      "1905 1905\n",
      "1906 1906\n",
      "1907 1907\n",
      "1908 1908\n",
      "1909 1909\n",
      "1910 1910\n",
      "1911 1911\n",
      "1912 1912\n",
      "1913 1913\n",
      "1914 1914\n",
      "1915 1915\n",
      "1916 1916\n",
      "1917 1917\n",
      "1918 1918\n",
      "1919 1919\n",
      "1920 1920\n",
      "1921 1921\n",
      "1922 1922\n",
      "1923 1923\n",
      "1924 1924\n",
      "1925 1925\n",
      "1926 1926\n",
      "1927 1927\n",
      "1928 1928\n",
      "1929 1929\n",
      "1930 1930\n",
      "1931 1931\n",
      "1932 1932\n",
      "1933 1933\n",
      "1934 1934\n",
      "1935 1935\n",
      "1936 1936\n",
      "1937 1937\n",
      "1938 1938\n",
      "1939 1939\n",
      "1940 1940\n",
      "1941 1941\n",
      "1942 1942\n",
      "1943 1943\n",
      "1944 1944\n",
      "1945 1945\n",
      "1946 1946\n",
      "1947 1947\n",
      "1948 1948\n",
      "1949 1949\n",
      "1950 1950\n",
      "1951 1951\n",
      "1952 1952\n",
      "1953 1953\n",
      "1954 1954\n",
      "1955 1955\n",
      "1956 1956\n",
      "1957 1957\n",
      "1958 1958\n",
      "1959 1959\n",
      "1960 1960\n",
      "1961 1961\n",
      "1962 1962\n",
      "1963 1963\n",
      "1964 1964\n",
      "1965 1965\n",
      "1966 1966\n",
      "1967 1967\n",
      "1968 1968\n",
      "1969 1969\n",
      "1970 1970\n",
      "1971 1971\n",
      "1972 1972\n",
      "1973 1973\n",
      "1974 1974\n",
      "1975 1975\n",
      "1976 1976\n",
      "1977 1977\n",
      "1978 1978\n",
      "1979 1979\n",
      "1980 1980\n",
      "1981 1981\n",
      "1982 1982\n",
      "1983 1983\n",
      "1984 1984\n",
      "1985 1985\n",
      "1986 1986\n",
      "1987 1987\n",
      "1988 1988\n",
      "1989 1989\n",
      "1990 1990\n",
      "1991 1991\n",
      "1992 1992\n",
      "1993 1993\n",
      "1994 1994\n",
      "1995 1995\n",
      "1996 1996\n",
      "1997 1997\n",
      "1998 1998\n",
      "1999 1999\n",
      "2000 2000\n",
      "2001 2001\n",
      "2002 2002\n",
      "2003 2003\n",
      "2004 2004\n",
      "2005 2005\n",
      "2006 2006\n",
      "2007 2007\n",
      "2008 2008\n",
      "2009 2009\n",
      "2010 2010\n",
      "2011 2011\n",
      "2012 2012\n",
      "2013 2013\n",
      "2014 2014\n",
      "2015 2015\n",
      "2016 2016\n",
      "2017 2017\n",
      "2018 2018\n",
      "2019 2019\n",
      "2020 2020\n",
      "2021 2021\n",
      "2022 2022\n",
      "2023 2023\n",
      "2024 2024\n",
      "2025 2025\n",
      "2026 2026\n",
      "2027 2027\n",
      "2028 2028\n",
      "2029 2029\n",
      "2030 2030\n",
      "2031 2031\n",
      "2032 2032\n",
      "2033 2033\n",
      "2034 2034\n",
      "2035 2035\n",
      "2036 2036\n",
      "2037 2037\n",
      "2038 2038\n",
      "2039 2039\n",
      "2040 2040\n",
      "2041 2041\n",
      "2042 2042\n",
      "2043 2043\n",
      "2044 2044\n",
      "2045 2045\n",
      "2046 2046\n",
      "2047 2047\n",
      "2048 2048\n",
      "2049 2049\n",
      "2050 2050\n",
      "2051 2051\n",
      "2052 2052\n",
      "2053 2053\n",
      "2054 2054\n",
      "2055 2055\n",
      "2056 2056\n",
      "2057 2057\n",
      "2058 2058\n",
      "2059 2059\n",
      "2060 2060\n",
      "2061 2061\n",
      "2062 2062\n",
      "2063 2063\n",
      "2064 2064\n",
      "2065 2065\n",
      "2066 2066\n",
      "2067 2067\n",
      "2068 2068\n",
      "2069 2069\n",
      "2070 2070\n",
      "2071 2071\n",
      "2072 2072\n",
      "2073 2073\n",
      "2074 2074\n",
      "2075 2075\n",
      "2076 2076\n",
      "2077 2077\n",
      "2078 2078\n",
      "2079 2079\n",
      "2080 2080\n",
      "2081 2081\n",
      "2082 2082\n",
      "2083 2083\n",
      "2084 2084\n",
      "2085 2085\n",
      "2086 2086\n",
      "2087 2087\n",
      "2088 2088\n",
      "2089 2089\n",
      "2090 2090\n",
      "2091 2091\n",
      "2092 2092\n",
      "2093 2093\n",
      "2094 2094\n",
      "2095 2095\n",
      "2096 2096\n",
      "2097 2097\n",
      "2098 2098\n",
      "2099 2099\n",
      "2100 2100\n",
      "2101 2101\n",
      "2102 2102\n",
      "2103 2103\n",
      "2104 2104\n",
      "2105 2105\n",
      "2106 2106\n",
      "2107 2107\n",
      "2108 2108\n",
      "2109 2109\n",
      "2110 2110\n",
      "2111 2111\n",
      "2112 2112\n",
      "2113 2113\n",
      "2114 2114\n",
      "2115 2115\n",
      "2116 2116\n",
      "2117 2117\n",
      "2118 2118\n",
      "2119 2119\n",
      "2120 2120\n",
      "2121 2121\n",
      "2122 2122\n",
      "2123 2123\n",
      "2124 2124\n",
      "2125 2125\n",
      "2126 2126\n",
      "2127 2127\n",
      "2128 2128\n",
      "2129 2129\n",
      "2130 2130\n",
      "2131 2131\n",
      "2132 2132\n",
      "2133 2133\n",
      "2134 2134\n",
      "2135 2135\n",
      "2136 2136\n",
      "2137 2137\n",
      "2138 2138\n",
      "2139 2139\n",
      "2140 2140\n",
      "2141 2141\n",
      "2142 2142\n",
      "2143 2143\n",
      "2144 2144\n",
      "2145 2145\n",
      "2146 2146\n",
      "2147 2147\n",
      "2148 2148\n",
      "2149 2149\n",
      "2150 2150\n",
      "2151 2151\n",
      "2152 2152\n",
      "2153 2153\n",
      "2154 2154\n",
      "2155 2155\n",
      "2156 2156\n",
      "2157 2157\n",
      "2158 2158\n",
      "2159 2159\n",
      "2160 2160\n",
      "2161 2161\n",
      "2162 2162\n",
      "2163 2163\n",
      "2164 2164\n",
      "2165 2165\n",
      "2166 2166\n",
      "2167 2167\n",
      "2168 2168\n",
      "2169 2169\n",
      "2170 2170\n",
      "2171 2171\n",
      "2172 2172\n",
      "2173 2173\n",
      "2174 2174\n",
      "2175 2175\n",
      "2176 2176\n",
      "2177 2177\n",
      "2178 2178\n",
      "2179 2179\n",
      "2180 2180\n",
      "2181 2181\n",
      "2182 2182\n",
      "2183 2183\n",
      "2184 2184\n",
      "2185 2185\n",
      "2186 2186\n",
      "2187 2187\n",
      "2188 2188\n",
      "2189 2189\n",
      "2190 2190\n",
      "2191 2191\n",
      "2192 2192\n",
      "2193 2193\n",
      "2194 2194\n",
      "2195 2195\n",
      "2196 2196\n",
      "2197 2197\n",
      "2198 2198\n",
      "2199 2199\n",
      "2200 2200\n",
      "2201 2201\n",
      "2202 2202\n",
      "2203 2203\n",
      "2204 2204\n",
      "2205 2205\n",
      "2206 2206\n",
      "2207 2207\n",
      "2208 2208\n",
      "2209 2209\n",
      "2210 2210\n",
      "2211 2211\n",
      "2212 2212\n",
      "2213 2213\n",
      "2214 2214\n",
      "2215 2215\n",
      "2216 2216\n",
      "2217 2217\n",
      "2218 2218\n",
      "2219 2219\n",
      "2220 2220\n",
      "2221 2221\n",
      "2222 2222\n",
      "2223 2223\n",
      "2224 2224\n",
      "2225 2225\n",
      "2226 2226\n",
      "2227 2227\n",
      "2228 2228\n",
      "2229 2229\n",
      "2230 2230\n",
      "2231 2231\n",
      "2232 2232\n",
      "2233 2233\n",
      "2234 2234\n",
      "2235 2235\n",
      "2236 2236\n",
      "2237 2237\n",
      "2238 2238\n",
      "2239 2239\n",
      "2240 2240\n",
      "2241 2241\n",
      "2242 2242\n",
      "2243 2243\n",
      "2244 2244\n",
      "2245 2245\n",
      "2246 2246\n",
      "2247 2247\n",
      "2248 2248\n",
      "2249 2249\n",
      "2250 2250\n",
      "2251 2251\n",
      "2252 2252\n",
      "2253 2253\n",
      "2254 2254\n",
      "2255 2255\n",
      "2256 2256\n",
      "2257 2257\n",
      "2258 2258\n",
      "2259 2259\n",
      "2260 2260\n",
      "2261 2261\n",
      "2262 2262\n",
      "2263 2263\n",
      "2264 2264\n",
      "2265 2265\n",
      "2266 2266\n",
      "2267 2267\n",
      "2268 2268\n",
      "2269 2269\n",
      "2270 2270\n",
      "2271 2271\n",
      "2272 2272\n",
      "2273 2273\n",
      "2274 2274\n",
      "2275 2275\n",
      "2276 2276\n",
      "2277 2277\n",
      "2278 2278\n",
      "2279 2279\n",
      "2280 2280\n",
      "2281 2281\n",
      "2282 2282\n",
      "2283 2283\n",
      "2284 2284\n",
      "2285 2285\n",
      "2286 2286\n",
      "2287 2287\n",
      "2288 2288\n",
      "2289 2289\n",
      "2290 2290\n",
      "2291 2291\n",
      "2292 2292\n",
      "2293 2293\n",
      "2294 2294\n",
      "2295 2295\n",
      "2296 2296\n",
      "2297 2297\n",
      "2298 2298\n",
      "2299 2299\n",
      "2300 2300\n",
      "2301 2301\n",
      "2302 2302\n",
      "2303 2303\n",
      "2304 2304\n",
      "2305 2305\n",
      "2306 2306\n",
      "2307 2307\n",
      "2308 2308\n",
      "2309 2309\n",
      "2310 2310\n",
      "2311 2311\n",
      "2312 2312\n",
      "2313 2313\n",
      "2314 2314\n",
      "2315 2315\n",
      "2316 2316\n",
      "2317 2317\n",
      "2318 2318\n",
      "2319 2319\n",
      "2320 2320\n",
      "2321 2321\n",
      "2322 2322\n",
      "2323 2323\n",
      "2324 2324\n",
      "2325 2325\n",
      "2326 2326\n",
      "2327 2327\n",
      "2328 2328\n",
      "2329 2329\n",
      "2330 2330\n",
      "2331 2331\n",
      "2332 2332\n",
      "2333 2333\n",
      "2334 2334\n",
      "2335 2335\n",
      "2336 2336\n",
      "2337 2337\n",
      "2338 2338\n",
      "2339 2339\n",
      "2340 2340\n",
      "2341 2341\n",
      "2342 2342\n",
      "2343 2343\n",
      "2344 2344\n",
      "2345 2345\n",
      "2346 2346\n",
      "2347 2347\n",
      "2348 2348\n",
      "2349 2349\n",
      "2350 2350\n",
      "2351 2351\n",
      "2352 2352\n",
      "2353 2353\n",
      "2354 2354\n",
      "2355 2355\n",
      "2356 2356\n",
      "2357 2357\n",
      "2358 2358\n",
      "2359 2359\n",
      "2360 2360\n",
      "2361 2361\n",
      "2362 2362\n",
      "2363 2363\n",
      "2364 2364\n",
      "2365 2365\n",
      "2366 2366\n",
      "2367 2367\n",
      "2368 2368\n",
      "2369 2369\n",
      "2370 2370\n",
      "2371 2371\n",
      "2372 2372\n",
      "2373 2373\n",
      "2374 2374\n",
      "2375 2375\n",
      "2376 2376\n",
      "2377 2377\n",
      "2378 2378\n",
      "2379 2379\n",
      "2380 2380\n",
      "2381 2381\n",
      "2382 2382\n",
      "2383 2383\n",
      "2384 2384\n",
      "2385 2385\n",
      "2386 2386\n",
      "2387 2387\n",
      "2388 2388\n",
      "2389 2389\n",
      "2390 2390\n",
      "2391 2391\n",
      "2392 2392\n",
      "2393 2393\n",
      "2394 2394\n",
      "2395 2395\n",
      "2396 2396\n",
      "2397 2397\n",
      "2398 2398\n",
      "2399 2399\n",
      "2400 2400\n",
      "2401 2401\n",
      "2402 2402\n",
      "2403 2403\n",
      "2404 2404\n",
      "2405 2405\n",
      "2406 2406\n",
      "2407 2407\n",
      "2408 2408\n",
      "2409 2409\n",
      "2410 2410\n",
      "2411 2411\n",
      "2412 2412\n",
      "2413 2413\n",
      "2414 2414\n",
      "2415 2415\n",
      "2416 2416\n",
      "2417 2417\n",
      "2418 2418\n",
      "2419 2419\n",
      "2420 2420\n",
      "2421 2421\n",
      "2422 2422\n",
      "2423 2423\n",
      "2424 2424\n",
      "2425 2425\n",
      "2426 2426\n",
      "2427 2427\n",
      "2428 2428\n",
      "2429 2429\n",
      "2430 2430\n",
      "2431 2431\n",
      "2432 2432\n",
      "2433 2433\n",
      "2434 2434\n",
      "2435 2435\n",
      "2436 2436\n",
      "2437 2437\n",
      "2438 2438\n",
      "2439 2439\n",
      "2440 2440\n",
      "2441 2441\n",
      "2442 2442\n",
      "2443 2443\n",
      "2444 2444\n",
      "2445 2445\n",
      "2446 2446\n",
      "2447 2447\n",
      "2448 2448\n",
      "2449 2449\n",
      "2450 2450\n",
      "2451 2451\n",
      "2452 2452\n",
      "2453 2453\n",
      "2454 2454\n",
      "2455 2455\n",
      "2456 2456\n",
      "2457 2457\n",
      "2458 2458\n",
      "2459 2459\n",
      "2460 2460\n",
      "2461 2461\n",
      "2462 2462\n",
      "2463 2463\n",
      "2464 2464\n",
      "2465 2465\n",
      "2466 2466\n",
      "2467 2467\n",
      "2468 2468\n",
      "2469 2469\n",
      "2470 2470\n",
      "2471 2471\n",
      "2472 2472\n",
      "2473 2473\n",
      "2474 2474\n",
      "2475 2475\n",
      "2476 2476\n",
      "2477 2477\n",
      "2478 2478\n",
      "2479 2479\n",
      "2480 2480\n",
      "2481 2481\n",
      "2482 2482\n",
      "2483 2483\n",
      "2484 2484\n",
      "2485 2485\n",
      "2486 2486\n",
      "2487 2487\n",
      "2488 2488\n",
      "2489 2489\n",
      "2490 2490\n",
      "2491 2491\n",
      "2492 2492\n",
      "2493 2493\n",
      "2494 2494\n",
      "2495 2495\n",
      "2496 2496\n",
      "2497 2497\n",
      "2498 2498\n",
      "2499 2499\n",
      "2500 2500\n",
      "2501 2501\n",
      "2502 2502\n",
      "2503 2503\n",
      "2504 2504\n",
      "2505 2505\n",
      "2506 2506\n",
      "2507 2507\n",
      "2508 2508\n",
      "2509 2509\n",
      "2510 2510\n",
      "2511 2511\n",
      "2512 2512\n",
      "2513 2513\n",
      "2514 2514\n",
      "2515 2515\n",
      "2516 2516\n",
      "2517 2517\n",
      "2518 2518\n",
      "2519 2519\n",
      "2520 2520\n",
      "2521 2521\n",
      "2522 2522\n",
      "2523 2523\n",
      "2524 2524\n",
      "2525 2525\n",
      "2526 2526\n",
      "2527 2527\n",
      "2528 2528\n",
      "2529 2529\n",
      "2530 2530\n",
      "2531 2531\n",
      "2532 2532\n",
      "2533 2533\n",
      "2534 2534\n",
      "2535 2535\n",
      "2536 2536\n",
      "2537 2537\n",
      "2538 2538\n",
      "2539 2539\n",
      "2540 2540\n",
      "2541 2541\n",
      "2542 2542\n",
      "2543 2543\n",
      "2544 2544\n",
      "2545 2545\n",
      "2546 2546\n",
      "2547 2547\n",
      "2548 2548\n",
      "2549 2549\n",
      "2550 2550\n",
      "2551 2551\n",
      "2552 2552\n",
      "2553 2553\n",
      "2554 2554\n",
      "2555 2555\n",
      "2556 2556\n",
      "2557 2557\n",
      "2558 2558\n",
      "2559 2559\n",
      "2560 2560\n",
      "2561 2561\n",
      "2562 2562\n",
      "2563 2563\n",
      "2564 2564\n",
      "2565 2565\n",
      "2566 2566\n",
      "2567 2567\n",
      "2568 2568\n",
      "2569 2569\n",
      "2570 2570\n",
      "2571 2571\n",
      "2572 2572\n",
      "2573 2573\n",
      "2574 2574\n",
      "2575 2575\n",
      "2576 2576\n",
      "2577 2577\n",
      "2578 2578\n",
      "2579 2579\n",
      "2580 2580\n",
      "2581 2581\n",
      "2582 2582\n",
      "2583 2583\n",
      "2584 2584\n",
      "2585 2585\n",
      "2586 2586\n",
      "2587 2587\n",
      "2588 2588\n",
      "2589 2589\n",
      "2590 2590\n",
      "2591 2591\n",
      "2592 2592\n",
      "2593 2593\n",
      "2594 2594\n",
      "2595 2595\n",
      "2596 2596\n",
      "2597 2597\n",
      "2598 2598\n",
      "2599 2599\n",
      "2600 2600\n",
      "2601 2601\n",
      "2602 2602\n",
      "2603 2603\n",
      "2604 2604\n",
      "2605 2605\n",
      "2606 2606\n",
      "2607 2607\n",
      "2608 2608\n",
      "2609 2609\n",
      "2610 2610\n",
      "2611 2611\n",
      "2612 2612\n",
      "2613 2613\n",
      "2614 2614\n",
      "2615 2615\n",
      "2616 2616\n",
      "2617 2617\n",
      "2618 2618\n",
      "2619 2619\n",
      "2620 2620\n",
      "2621 2621\n",
      "2622 2622\n",
      "2623 2623\n",
      "2624 2624\n",
      "2625 2625\n",
      "2626 2626\n",
      "2627 2627\n",
      "2628 2628\n",
      "2629 2629\n",
      "2630 2630\n",
      "2631 2631\n",
      "2632 2632\n",
      "2633 2633\n",
      "2634 2634\n",
      "2635 2635\n",
      "2636 2636\n",
      "2637 2637\n",
      "2638 2638\n",
      "2639 2639\n",
      "2640 2640\n",
      "2641 2641\n",
      "2642 2642\n",
      "2643 2643\n",
      "2644 2644\n",
      "2645 2645\n",
      "2646 2646\n",
      "2647 2647\n",
      "2648 2648\n",
      "2649 2649\n",
      "2650 2650\n",
      "2651 2651\n",
      "2652 2652\n",
      "2653 2653\n",
      "2654 2654\n",
      "2655 2655\n",
      "2656 2656\n",
      "2657 2657\n",
      "2658 2658\n",
      "2659 2659\n",
      "2660 2660\n",
      "2661 2661\n",
      "2662 2662\n",
      "2663 2663\n",
      "2664 2664\n",
      "2665 2665\n",
      "2666 2666\n",
      "2667 2667\n",
      "2668 2668\n",
      "2669 2669\n",
      "2670 2670\n",
      "2671 2671\n",
      "2672 2672\n",
      "2673 2673\n",
      "2674 2674\n",
      "2675 2675\n",
      "2676 2676\n",
      "2677 2677\n",
      "2678 2678\n",
      "2679 2679\n",
      "2680 2680\n",
      "2681 2681\n",
      "2682 2682\n",
      "2683 2683\n",
      "2684 2684\n",
      "2685 2685\n",
      "2686 2686\n",
      "2687 2687\n",
      "2688 2688\n",
      "2689 2689\n",
      "2690 2690\n",
      "2691 2691\n",
      "2692 2692\n",
      "2693 2693\n",
      "2694 2694\n",
      "2695 2695\n",
      "2696 2696\n",
      "2697 2697\n",
      "2698 2698\n",
      "2699 2699\n",
      "2700 2700\n",
      "2701 2701\n",
      "2702 2702\n",
      "2703 2703\n",
      "2704 2704\n",
      "2705 2705\n",
      "2706 2706\n",
      "2707 2707\n",
      "2708 2708\n",
      "2709 2709\n",
      "2710 2710\n",
      "2711 2711\n",
      "2712 2712\n",
      "2713 2713\n",
      "2714 2714\n",
      "2715 2715\n",
      "2716 2716\n",
      "2717 2717\n",
      "2718 2718\n",
      "2719 2719\n",
      "2720 2720\n",
      "2721 2721\n",
      "2722 2722\n",
      "2723 2723\n",
      "2724 2724\n",
      "2725 2725\n",
      "2726 2726\n"
     ]
    }
   ],
   "source": [
    "for ii in range(len(spindle_training_data)):\n",
    "    spindle_training_data.append(non_spindle_training_data[ii])\n",
    "    spindle_label.append(non_spindle_label[ii])\n",
    "    print(len(spindle_training_data),len(spindle_label))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "training_data_temp = spindle_training_data\n",
    "training_label_temp = spindle_label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(2720, 18000) (2720,)\n"
     ]
    }
   ],
   "source": [
    "training_data=[];training_label=[]\n",
    "for items in zip(training_data_temp,training_label_temp):\n",
    "    if len(items[0]) == 18000:\n",
    "        training_data.append(items[0])\n",
    "        training_label.append(items[1])\n",
    "print(np.array(training_data).shape,np.array(training_label).shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1, 1, 1, 1, 1]\n",
      "[1, 1, 1, 1, 1]\n"
     ]
    }
   ],
   "source": [
    "print(training_label[:5])\n",
    "np.random.shuffle(training_data)\n",
    "np.random.shuffle(training_label)\n",
    "print(training_label[:5])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(training_data, training_label, test_size=0.20, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "             precision    recall  f1-score   support\n",
      "\n",
      "          0       0.32      0.55      0.40       176\n",
      "          1       0.67      0.45      0.54       368\n",
      "\n",
      "avg / total       0.56      0.48      0.49       544\n",
      "\n",
      "0.543658088235\n"
     ]
    }
   ],
   "source": [
    "#clf = OneVsRestClassifier(SVC(kernel='linear',tol=1e-6,max_iter=5e5))\n",
    "clf = OneVsRestClassifier(LogisticRegression(penalty='l2',C=5,tol=10e-9,fit_intercept=True,solver='liblinear',\n",
    "                                             max_iter=5e8,multi_class='ovr',n_jobs=-1))\n",
    "\n",
    "\n",
    "\n",
    "clf.fit(X_train,y_train)\n",
    "print(classification_report(clf.predict(X_test),y_test))\n",
    "print(clf.score(X_train,y_train))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ 0.50612745  0.51960784  0.46078431]\n"
     ]
    }
   ],
   "source": [
    "classification = make_pipeline(StandardScaler(),clf)\n",
    "n_samples=np.array(training_data).shape[0]\n",
    "cv = ShuffleSplit(n_samples, n_iter=3,test_size=0.3, random_state=0)\n",
    "scores = cross_val_score(classification, np.array(training_data),np.array(training_label),cv=cv, scoring='accuracy')\n",
    "\n",
    "print(scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "             precision    recall  f1-score   support\n",
      "\n",
      "          0       0.50      0.57      0.53       265\n",
      "          1       0.53      0.47      0.50       279\n",
      "\n",
      "avg / total       0.52      0.52      0.52       544\n",
      "\n",
      "0.929227941176\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Anaconda3\\lib\\site-packages\\sklearn\\discriminant_analysis.py:387: UserWarning: Variables are collinear.\n",
      "  warnings.warn(\"Variables are collinear.\")\n"
     ]
    }
   ],
   "source": [
    "clf = LDA(solver='svd', shrinkage=None, priors=None, n_components=100, store_covariance=False, tol=10e-8)\n",
    "clf.fit(X_train,y_train)\n",
    "print(classification_report(clf.predict(X_test),y_test))\n",
    "print(clf.score(X_train,y_train))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Anaconda3\\lib\\site-packages\\sklearn\\discriminant_analysis.py:387: UserWarning: Variables are collinear.\n",
      "  warnings.warn(\"Variables are collinear.\")\n",
      "C:\\Anaconda3\\lib\\site-packages\\sklearn\\discriminant_analysis.py:387: UserWarning: Variables are collinear.\n",
      "  warnings.warn(\"Variables are collinear.\")\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ 0.48897059  0.52573529  0.49264706]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Anaconda3\\lib\\site-packages\\sklearn\\discriminant_analysis.py:387: UserWarning: Variables are collinear.\n",
      "  warnings.warn(\"Variables are collinear.\")\n"
     ]
    }
   ],
   "source": [
    "classification = make_pipeline(StandardScaler(),clf)\n",
    "n_samples=np.array(training_data).shape[0]\n",
    "cv = ShuffleSplit(n_samples, n_iter=3,test_size=0.3, random_state=0)\n",
    "scores = cross_val_score(classification, np.array(training_data),np.array(training_label),cv=cv, scoring='accuracy')\n",
    "\n",
    "print(scores)"
   ]
  },
  {
   "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": "3.5.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}