CNN_injection_superposition_4Chann.ipynb 77.5 KB
Newer Older
Bannier Delphine's avatar
Bannier Delphine committed
1
2
3
4
5
6
7
8
9
10
11
12
13
{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# CNN collage + MLP"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
Bannier Delphine's avatar
Bannier Delphine committed
14
15
16
    "https://www.pyimagesearch.com/2019/02/04/keras-multiple-inputs-and-mixed-data/\n",
    "\n",
    "https://www.kaggle.com/franklemuchahary/basic-cnn-keras-with-cross-validation"
Bannier Delphine's avatar
Bannier Delphine committed
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import os\n",
    "import logging\n",
    "logging.basicConfig(format='%(levelname)s:%(message)s', level=logging.INFO)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## A - Preprocessing : Reading Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "os.chdir('../')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Patient</th>\n",
       "      <th>Weeks</th>\n",
       "      <th>FVC</th>\n",
       "      <th>Percent</th>\n",
       "      <th>Age</th>\n",
       "      <th>Sex</th>\n",
       "      <th>SmokingStatus</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ID00007637202177411956430</td>\n",
       "      <td>-4</td>\n",
       "      <td>2315</td>\n",
       "      <td>58.253649</td>\n",
       "      <td>79</td>\n",
       "      <td>Male</td>\n",
       "      <td>Ex-smoker</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ID00007637202177411956430</td>\n",
       "      <td>5</td>\n",
       "      <td>2214</td>\n",
       "      <td>55.712129</td>\n",
       "      <td>79</td>\n",
       "      <td>Male</td>\n",
       "      <td>Ex-smoker</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>ID00007637202177411956430</td>\n",
       "      <td>7</td>\n",
       "      <td>2061</td>\n",
       "      <td>51.862104</td>\n",
       "      <td>79</td>\n",
       "      <td>Male</td>\n",
       "      <td>Ex-smoker</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>ID00007637202177411956430</td>\n",
       "      <td>9</td>\n",
       "      <td>2144</td>\n",
       "      <td>53.950679</td>\n",
       "      <td>79</td>\n",
       "      <td>Male</td>\n",
       "      <td>Ex-smoker</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>ID00007637202177411956430</td>\n",
       "      <td>11</td>\n",
       "      <td>2069</td>\n",
       "      <td>52.063412</td>\n",
       "      <td>79</td>\n",
       "      <td>Male</td>\n",
       "      <td>Ex-smoker</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                     Patient  Weeks   FVC    Percent  Age   Sex SmokingStatus\n",
       "0  ID00007637202177411956430     -4  2315  58.253649   79  Male     Ex-smoker\n",
       "1  ID00007637202177411956430      5  2214  55.712129   79  Male     Ex-smoker\n",
       "2  ID00007637202177411956430      7  2061  51.862104   79  Male     Ex-smoker\n",
       "3  ID00007637202177411956430      9  2144  53.950679   79  Male     Ex-smoker\n",
       "4  ID00007637202177411956430     11  2069  52.063412   79  Male     Ex-smoker"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from preprocessing.read_load_data import read_data\n",
    "\n",
    "input_directory='../osic-pulmonary-fibrosis-progression'\n",
    "train_df, test_df, sample_df = read_data(input_directory)   \n",
    "train_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## B - Preprocessing : Loading Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "patients_train_ids= train_df.Patient.unique()\n",
    "patient_test_list= test_df.Patient.unique()\n",
    "patients_train_ids = [pat for pat in patients_train_ids]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:loading  attributes...\n",
      "INFO:loading images...\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Array shape:  (176, 240, 240)\n",
      "min value:  -0.124956130312047\n",
      "max value:  0.13136708968066257\n"
     ]
    }
   ],
   "source": [
    "from preprocessing.read_load_data import load_images\n",
    "\n",
    "logging.info(\"loading  attributes...\")\n",
    "df = pd.read_csv(f'{input_directory}/train.csv')\n",
    "patients_train_ids= df.Patient.unique().tolist()\n",
    "\n",
    "logging.info(\"loading images...\")\n",
    "images = load_images(input_directory,\n",
    "                    'train',\n",
    "                     patients_train_ids,\n",
    "                     option='collage',\n",
    "                     outputH = 240,\n",
    "                     outputW = 240)\n",
    "\n",
    "print(\"Array shape: \", images.shape)\n",
    "#check value between -1,1\n",
    "print('min value: ', np.amin(images))\n",
    "print('max value: ', np.amax(images))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## C - Preprocessing : shuffle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "\n",
Bannier Delphine's avatar
Bannier Delphine committed
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
    "split = train_test_split(patients_train_ids, images, test_size=0.2, random_state=42)\n",
    "(trainPatient, testPatient, trainImagesX, testImagesX) = split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "#split the dataframe like the images\n",
    "df_train = df[df.Patient.isin(trainPatient)].copy()\n",
    "df_test = df[df.Patient.isin(testPatient)].copy()"
   ]
  },
  {
   "cell_type": "code",
Bannier Delphine's avatar
Bannier Delphine committed
254
   "execution_count": 8,
Bannier Delphine's avatar
Bannier Delphine committed
255
   "metadata": {},
256
   "outputs": [
Bannier Delphine's avatar
Bannier Delphine committed
257
258
259
260
261
262
263
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:NumExpr defaulting to 8 threads.\n"
     ]
    },
264
265
266
267
268
269
270
271
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1093, 8) (280, 8)\n"
     ]
    }
   ],
Bannier Delphine's avatar
Bannier Delphine committed
272
273
274
275
   "source": [
    "from preprocessing.read_load_data import create_dataframe\n",
    "\n",
    "trainAttrX = create_dataframe(df_train)\n",
276
277
278
279
280
281
    "testAttrX = create_dataframe(df_test)\n",
    "print(trainAttrX.shape, testAttrX.shape)"
   ]
  },
  {
   "cell_type": "code",
Bannier Delphine's avatar
Bannier Delphine committed
282
   "execution_count": 9,
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1093 280\n"
     ]
    }
   ],
   "source": [
    "#set one image per training row\n",
    "\n",
    "indice = 0\n",
    "train_dataset = np.ndarray((len(trainAttrX),240,240))\n",
Bannier Delphine's avatar
Bannier Delphine committed
298
    "for i,patient in enumerate(trainPatient):\n",
299
300
301
302
303
304
305
306
    "    nb_data = len(trainAttrX[trainAttrX.PatientID ==patient])\n",
    "    for ii in range(nb_data):\n",
    "        train_dataset[indice]=(trainImagesX[i])\n",
    "        indice+=1\n",
    "        \n",
    "        \n",
    "indicet = 0        \n",
    "test_dataset = np.ndarray((len(testAttrX),240,240))\n",
Bannier Delphine's avatar
Bannier Delphine committed
307
    "for i,patient in enumerate(testPatient):\n",
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
    "    nb_data = len(testAttrX[testAttrX.PatientID ==patient])\n",
    "    for ii in range(nb_data):\n",
    "        test_dataset[indicet] = testImagesX[i]\n",
    "        indicet+=1\n",
    "        \n",
    "        \n",
    "print(len(train_dataset),len(test_dataset))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## D - Preprocessing : Scaling + Encoding"
   ]
  },
  {
   "cell_type": "code",
Bannier Delphine's avatar
Bannier Delphine committed
326
   "execution_count": 10,
327
328
329
330
331
332
   "metadata": {},
   "outputs": [],
   "source": [
    "from preprocessing.scale_data import scale_variable\n",
    "\n",
    "sc, trainAttrX, testAttrX = scale_variable(trainAttrX, testAttrX,'Target_FVC')\n",
Bannier Delphine's avatar
Bannier Delphine committed
333
334
335
    "sc1, trainAttrX, testAttrX = scale_variable(trainAttrX, testAttrX,'First_FVC')\n",
    "sc2, trainAttrX, testAttrX = scale_variable(trainAttrX, testAttrX,'Age')\n",
    "\n",
336
337
338
339
340
341
    "trainY = trainAttrX.loc[:,'Target_FVC_scaled']\n",
    "testY = testAttrX.loc[:,'Target_FVC_scaled']"
   ]
  },
  {
   "cell_type": "code",
Bannier Delphine's avatar
Bannier Delphine committed
342
   "execution_count": 11,
343
344
345
346
347
348
349
350
   "metadata": {},
   "outputs": [],
   "source": [
    "from preprocessing.scale_data import encode_variable\n",
    "\n",
    "trainAttrX, testAttrX = encode_variable(trainAttrX, testAttrX,'Sex')\n",
    "trainAttrX, testAttrX = encode_variable(trainAttrX, testAttrX,'SmokingStatus')\n",
    "\n",
Bannier Delphine's avatar
Bannier Delphine committed
351
352
353
354
355
    "for dft in [trainAttrX,testAttrX]:\n",
    "    dft.drop(columns = ['Sex','SmokingStatus','Target_FVC','Target_FVC_scaled',\n",
    "                          'PatientID','First_FVC','Age'], inplace = True)\n",
    "    dft.loc[:,'First_Percent'] = dft.loc[:,'First_Percent']/100\n",
    "    dft.loc[:,'Delta_week'] = dft.loc[:,'Delta_week']/133"
Bannier Delphine's avatar
Bannier Delphine committed
356
357
358
359
   ]
  },
  {
   "cell_type": "code",
Bannier Delphine's avatar
Bannier Delphine committed
360
   "execution_count": 12,
Bannier Delphine's avatar
Bannier Delphine committed
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>First_Percent</th>\n",
       "      <th>Delta_week</th>\n",
Bannier Delphine's avatar
Bannier Delphine committed
386
387
       "      <th>First_FVC_scaled</th>\n",
       "      <th>Age_scaled</th>\n",
388
389
       "      <th>Sex_le</th>\n",
       "      <th>SmokingStatus_le</th>\n",
Bannier Delphine's avatar
Bannier Delphine committed
390
391
392
393
394
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
Bannier Delphine's avatar
Bannier Delphine committed
395
396
397
398
       "      <td>0.582536</td>\n",
       "      <td>0.067669</td>\n",
       "      <td>-0.631784</td>\n",
       "      <td>1.684379</td>\n",
399
400
       "      <td>1</td>\n",
       "      <td>1</td>\n",
Bannier Delphine's avatar
Bannier Delphine committed
401
402
403
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
Bannier Delphine's avatar
Bannier Delphine committed
404
405
406
407
       "      <td>0.582536</td>\n",
       "      <td>0.082707</td>\n",
       "      <td>-0.631784</td>\n",
       "      <td>1.684379</td>\n",
408
409
       "      <td>1</td>\n",
       "      <td>1</td>\n",
Bannier Delphine's avatar
Bannier Delphine committed
410
411
412
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
Bannier Delphine's avatar
Bannier Delphine committed
413
414
415
416
       "      <td>0.582536</td>\n",
       "      <td>0.097744</td>\n",
       "      <td>-0.631784</td>\n",
       "      <td>1.684379</td>\n",
417
418
       "      <td>1</td>\n",
       "      <td>1</td>\n",
Bannier Delphine's avatar
Bannier Delphine committed
419
420
421
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
Bannier Delphine's avatar
Bannier Delphine committed
422
423
424
425
       "      <td>0.582536</td>\n",
       "      <td>0.112782</td>\n",
       "      <td>-0.631784</td>\n",
       "      <td>1.684379</td>\n",
426
427
       "      <td>1</td>\n",
       "      <td>1</td>\n",
Bannier Delphine's avatar
Bannier Delphine committed
428
429
430
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
Bannier Delphine's avatar
Bannier Delphine committed
431
432
433
434
       "      <td>0.582536</td>\n",
       "      <td>0.157895</td>\n",
       "      <td>-0.631784</td>\n",
       "      <td>1.684379</td>\n",
435
436
       "      <td>1</td>\n",
       "      <td>1</td>\n",
Bannier Delphine's avatar
Bannier Delphine committed
437
438
439
440
441
442
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
Bannier Delphine's avatar
Bannier Delphine committed
443
444
445
446
447
448
449
450
451
452
453
454
455
       "   First_Percent  Delta_week  First_FVC_scaled  Age_scaled  Sex_le  \\\n",
       "0       0.582536    0.067669         -0.631784    1.684379       1   \n",
       "1       0.582536    0.082707         -0.631784    1.684379       1   \n",
       "2       0.582536    0.097744         -0.631784    1.684379       1   \n",
       "3       0.582536    0.112782         -0.631784    1.684379       1   \n",
       "4       0.582536    0.157895         -0.631784    1.684379       1   \n",
       "\n",
       "   SmokingStatus_le  \n",
       "0                 1  \n",
       "1                 1  \n",
       "2                 1  \n",
       "3                 1  \n",
       "4                 1  "
Bannier Delphine's avatar
Bannier Delphine committed
456
457
      ]
     },
Bannier Delphine's avatar
Bannier Delphine committed
458
     "execution_count": 12,
Bannier Delphine's avatar
Bannier Delphine committed
459
460
461
462
463
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
464
    "trainAttrX.head(5)"
Bannier Delphine's avatar
Bannier Delphine committed
465
466
467
468
469
470
471
472
473
474
475
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## E - Processing : Create models"
   ]
  },
  {
   "cell_type": "code",
Bannier Delphine's avatar
Bannier Delphine committed
476
   "execution_count": 13,
Bannier Delphine's avatar
Bannier Delphine committed
477
478
479
480
481
482
483
484
485
486
487
488
489
   "metadata": {},
   "outputs": [],
   "source": [
    "from processing.models import create_hybrid\n",
    "from keras.optimizers import Adam\n",
    "\n",
    "model = create_hybrid(trainAttrX.shape[1], shape = (240,240,1))\n",
    "opt = Adam(lr=1e-3, decay=1e-3 / 200)\n",
    "model.compile(loss=\"mean_absolute_percentage_error\", optimizer=opt)"
   ]
  },
  {
   "cell_type": "code",
Bannier Delphine's avatar
Bannier Delphine committed
490
   "execution_count": 14,
Bannier Delphine's avatar
Bannier Delphine committed
491
492
493
494
495
496
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
Bannier Delphine's avatar
Bannier Delphine committed
497
      "Model: \"model_1\"\n",
Bannier Delphine's avatar
Bannier Delphine committed
498
499
500
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
Bannier Delphine's avatar
Bannier Delphine committed
501
      "input_1 (InputLayer)            [(None, 240, 240, 1) 0                                            \n",
Bannier Delphine's avatar
Bannier Delphine committed
502
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
503
      "conv2d (Conv2D)                 (None, 240, 240, 32) 320         input_1[0][0]                    \n",
Bannier Delphine's avatar
Bannier Delphine committed
504
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
505
      "activation (Activation)         (None, 240, 240, 32) 0           conv2d[0][0]                     \n",
Bannier Delphine's avatar
Bannier Delphine committed
506
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
507
      "batch_normalization (BatchNorma (None, 240, 240, 32) 128         activation[0][0]                 \n",
Bannier Delphine's avatar
Bannier Delphine committed
508
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
509
      "max_pooling2d (MaxPooling2D)    (None, 120, 120, 32) 0           batch_normalization[0][0]        \n",
Bannier Delphine's avatar
Bannier Delphine committed
510
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
511
      "conv2d_1 (Conv2D)               (None, 120, 120, 64) 18496       max_pooling2d[0][0]              \n",
Bannier Delphine's avatar
Bannier Delphine committed
512
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
513
      "activation_1 (Activation)       (None, 120, 120, 64) 0           conv2d_1[0][0]                   \n",
Bannier Delphine's avatar
Bannier Delphine committed
514
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
515
      "batch_normalization_1 (BatchNor (None, 120, 120, 64) 256         activation_1[0][0]               \n",
Bannier Delphine's avatar
Bannier Delphine committed
516
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
517
      "max_pooling2d_1 (MaxPooling2D)  (None, 60, 60, 64)   0           batch_normalization_1[0][0]      \n",
Bannier Delphine's avatar
Bannier Delphine committed
518
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
519
      "conv2d_2 (Conv2D)               (None, 60, 60, 128)  73856       max_pooling2d_1[0][0]            \n",
Bannier Delphine's avatar
Bannier Delphine committed
520
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
521
      "activation_2 (Activation)       (None, 60, 60, 128)  0           conv2d_2[0][0]                   \n",
Bannier Delphine's avatar
Bannier Delphine committed
522
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
523
      "batch_normalization_2 (BatchNor (None, 60, 60, 128)  512         activation_2[0][0]               \n",
Bannier Delphine's avatar
Bannier Delphine committed
524
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
525
      "max_pooling2d_2 (MaxPooling2D)  (None, 30, 30, 128)  0           batch_normalization_2[0][0]      \n",
Bannier Delphine's avatar
Bannier Delphine committed
526
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
527
      "flatten (Flatten)               (None, 115200)       0           max_pooling2d_2[0][0]            \n",
Bannier Delphine's avatar
Bannier Delphine committed
528
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
529
      "dense_3 (Dense)                 (None, 16)           1843216     flatten[0][0]                    \n",
Bannier Delphine's avatar
Bannier Delphine committed
530
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
531
      "activation_3 (Activation)       (None, 16)           0           dense_3[0][0]                    \n",
Bannier Delphine's avatar
Bannier Delphine committed
532
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
533
      "dense_input (InputLayer)        [(None, 6)]          0                                            \n",
Bannier Delphine's avatar
Bannier Delphine committed
534
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
535
      "batch_normalization_3 (BatchNor (None, 16)           64          activation_3[0][0]               \n",
Bannier Delphine's avatar
Bannier Delphine committed
536
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
537
      "dense (Dense)                   (None, 8)            56          dense_input[0][0]                \n",
Bannier Delphine's avatar
Bannier Delphine committed
538
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
539
      "dropout (Dropout)               (None, 16)           0           batch_normalization_3[0][0]      \n",
Bannier Delphine's avatar
Bannier Delphine committed
540
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
541
      "dense_1 (Dense)                 (None, 4)            36          dense[0][0]                      \n",
Bannier Delphine's avatar
Bannier Delphine committed
542
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
543
      "dense_4 (Dense)                 (None, 4)            68          dropout[0][0]                    \n",
Bannier Delphine's avatar
Bannier Delphine committed
544
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
545
      "dense_2 (Dense)                 (None, 1)            5           dense_1[0][0]                    \n",
Bannier Delphine's avatar
Bannier Delphine committed
546
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
547
      "activation_4 (Activation)       (None, 4)            0           dense_4[0][0]                    \n",
Bannier Delphine's avatar
Bannier Delphine committed
548
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
549
550
      "concatenate (Concatenate)       (None, 5)            0           dense_2[0][0]                    \n",
      "                                                                 activation_4[0][0]               \n",
Bannier Delphine's avatar
Bannier Delphine committed
551
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
552
      "dense_5 (Dense)                 (None, 4)            24          concatenate[0][0]                \n",
Bannier Delphine's avatar
Bannier Delphine committed
553
      "__________________________________________________________________________________________________\n",
Bannier Delphine's avatar
Bannier Delphine committed
554
      "dense_6 (Dense)                 (None, 1)            5           dense_5[0][0]                    \n",
Bannier Delphine's avatar
Bannier Delphine committed
555
      "==================================================================================================\n",
Bannier Delphine's avatar
Bannier Delphine committed
556
557
      "Total params: 1,937,042\n",
      "Trainable params: 1,936,562\n",
Bannier Delphine's avatar
Bannier Delphine committed
558
559
560
561
562
563
564
565
566
567
568
      "Non-trainable params: 480\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
Bannier Delphine's avatar
Bannier Delphine committed
569
   "execution_count": 15,
Bannier Delphine's avatar
Bannier Delphine committed
570
   "metadata": {},
Bannier Delphine's avatar
Bannier Delphine committed
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
   "outputs": [],
   "source": [
    "from tensorflow.keras.callbacks import ModelCheckpoint, EarlyStopping\n",
    "\n",
    "#set early stopping criteria\n",
    "pat = 5 #this is the number of epochs with no improvment after which the training will stop\n",
    "es = EarlyStopping(monitor='val_loss', patience=pat, verbose=1)\n",
    "\n",
    "#define the model checkpoint callback -> this will keep on saving the model as a physical file\n",
    "cp = ModelCheckpoint('superposition_injection.h5', verbose=1, save_best_only=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "def custom_shuffle_split(trainAttrX,train_dataset,trainY,test_size = 0.1 ):\n",
    "    cut = int(len(trainY)*test_size)\n",
    "    arr = list(np.arange(len(trainY)))\n",
    "    np.random.shuffle(arr)\n",
    "    trainidx = arr[cut:]\n",
    "    testidx = arr[:cut]\n",
    "    train_x, train_y = [trainAttrX.iloc[trainidx], train_dataset[trainidx]] , trainY[trainidx]\n",
    "    val_x, val_y = [trainAttrX.iloc[testidx], train_dataset[testidx]] , trainY[testidx]\n",
    "    return train_x, val_x, train_y, val_y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
Bannier Delphine's avatar
Bannier Delphine committed
604
605
   "outputs": [
    {
606
607
608
     "name": "stdout",
     "output_type": "stream",
     "text": [
Bannier Delphine's avatar
Bannier Delphine committed
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
      "Training on Fold:  1\n",
      "Epoch 1/30\n",
      "111/111 [==============================] - ETA: 0s - loss: 316.5363\n",
      "Epoch 00001: val_loss improved from inf to 107.96893, saving model to superposition_injection.h5\n",
      "111/111 [==============================] - 56s 503ms/step - loss: 316.5363 - val_loss: 107.9689\n",
      "Epoch 2/30\n",
      "111/111 [==============================] - ETA: 0s - loss: 197.5600\n",
      "Epoch 00002: val_loss improved from 107.96893 to 100.10335, saving model to superposition_injection.h5\n",
      "111/111 [==============================] - 55s 495ms/step - loss: 197.5600 - val_loss: 100.1033\n",
      "Epoch 3/30\n",
      "111/111 [==============================] - ETA: 0s - loss: 220.9793\n",
      "Epoch 00003: val_loss improved from 100.10335 to 99.90191, saving model to superposition_injection.h5\n",
      "111/111 [==============================] - 54s 485ms/step - loss: 220.9793 - val_loss: 99.9019\n",
      "Epoch 4/30\n",
      "111/111 [==============================] - ETA: 0s - loss: 126.4834\n",
      "Epoch 00004: val_loss did not improve from 99.90191\n",
      "111/111 [==============================] - 55s 498ms/step - loss: 126.4834 - val_loss: 100.0706\n",
      "Epoch 5/30\n",
      "111/111 [==============================] - ETA: 0s - loss: 130.7720\n",
      "Epoch 00005: val_loss improved from 99.90191 to 99.70051, saving model to superposition_injection.h5\n",
      "111/111 [==============================] - 54s 485ms/step - loss: 130.7720 - val_loss: 99.7005\n",
      "Epoch 6/30\n",
      "111/111 [==============================] - ETA: 0s - loss: 164.4418\n",
      "Epoch 00006: val_loss did not improve from 99.70051\n",
      "111/111 [==============================] - 54s 483ms/step - loss: 164.4418 - val_loss: 100.6682\n",
      "Epoch 7/30\n",
      "111/111 [==============================] - ETA: 0s - loss: 119.5080\n",
      "Epoch 00007: val_loss did not improve from 99.70051\n",
      "111/111 [==============================] - 55s 497ms/step - loss: 119.5080 - val_loss: 101.0239\n",
      "Epoch 8/30\n",
      "111/111 [==============================] - ETA: 0s - loss: 151.3392\n",
      "Epoch 00008: val_loss did not improve from 99.70051\n",
      "111/111 [==============================] - 56s 505ms/step - loss: 151.3392 - val_loss: 100.1579\n",
      "Epoch 9/30\n",
      "111/111 [==============================] - ETA: 0s - loss: 105.6256\n",
      "Epoch 00009: val_loss did not improve from 99.70051\n",
      "111/111 [==============================] - 55s 492ms/step - loss: 105.6256 - val_loss: 100.2133\n",
      "Epoch 10/30\n",
      "111/111 [==============================] - ETA: 0s - loss: 114.8794\n",
      "Epoch 00010: val_loss did not improve from 99.70051\n",
      "111/111 [==============================] - 54s 491ms/step - loss: 114.8794 - val_loss: 100.3191\n",
      "Epoch 00010: early stopping\n",
      "4/4 [==============================] - 1s 200ms/step - loss: 100.1156\n",
      "Val Score:  100.11563110351562\n",
      "====================================================================================\n",
      "\n",
      "\n",
      "Training on Fold:  2\n",
      "Epoch 1/30\n",
      "111/111 [==============================] - ETA: 0s - loss: 168.3844\n",
      "Epoch 00001: val_loss did not improve from 99.70051\n",
      "111/111 [==============================] - 55s 494ms/step - loss: 168.3844 - val_loss: 107.7039\n",
      "Epoch 2/30\n",
      "111/111 [==============================] - ETA: 0s - loss: 111.5110\n",
      "Epoch 00002: val_loss did not improve from 99.70051\n",
      "111/111 [==============================] - 56s 507ms/step - loss: 111.5110 - val_loss: 109.4569\n",
      "Epoch 3/30\n",
      "111/111 [==============================] - ETA: 0s - loss: 114.4467\n",
      "Epoch 00003: val_loss did not improve from 99.70051\n",
      "111/111 [==============================] - 56s 505ms/step - loss: 114.4467 - val_loss: 101.8173\n",
      "Epoch 4/30\n",
      "111/111 [==============================] - ETA: 0s - loss: 112.2941\n",
      "Epoch 00004: val_loss did not improve from 99.70051\n",
      "111/111 [==============================] - 55s 492ms/step - loss: 112.2941 - val_loss: 109.3902\n",
      "Epoch 5/30\n",
      "111/111 [==============================] - ETA: 0s - loss: 112.2740\n",
      "Epoch 00005: val_loss did not improve from 99.70051\n",
      "111/111 [==============================] - 56s 503ms/step - loss: 112.2740 - val_loss: 101.4342\n",
      "Epoch 6/30\n",
      "111/111 [==============================] - ETA: 0s - loss: 111.4760\n",
      "Epoch 00006: val_loss did not improve from 99.70051\n",
      "111/111 [==============================] - 55s 500ms/step - loss: 111.4760 - val_loss: 99.8436\n",
      "Epoch 7/30\n",
      "111/111 [==============================] - ETA: 0s - loss: 110.4311\n",
      "Epoch 00007: val_loss did not improve from 99.70051\n",
      "111/111 [==============================] - 56s 503ms/step - loss: 110.4311 - val_loss: 110.9042\n",
      "Epoch 8/30\n",
      "111/111 [==============================] - ETA: 0s - loss: 107.2995\n",
      "Epoch 00008: val_loss did not improve from 99.70051\n",
      "111/111 [==============================] - 58s 520ms/step - loss: 107.2995 - val_loss: 100.7832\n",
      "Epoch 9/30\n",
      "111/111 [==============================] - ETA: 0s - loss: 103.8186\n",
      "Epoch 00009: val_loss did not improve from 99.70051\n",
      "111/111 [==============================] - 56s 500ms/step - loss: 103.8186 - val_loss: 103.2870\n",
      "Epoch 10/30\n",
      "111/111 [==============================] - ETA: 0s - loss: 101.8159\n",
      "Epoch 00010: val_loss did not improve from 99.70051\n",
      "111/111 [==============================] - 54s 489ms/step - loss: 101.8159 - val_loss: 100.8220\n",
      "Epoch 11/30\n",
      "111/111 [==============================] - ETA: 0s - loss: 104.4015\n",
      "Epoch 00011: val_loss did not improve from 99.70051\n",
      "111/111 [==============================] - 54s 483ms/step - loss: 104.4015 - val_loss: 110.9032\n",
      "Epoch 00011: early stopping\n",
      "4/4 [==============================] - 1s 219ms/step - loss: 108.3460\n",
      "Val Score:  108.34602355957031\n",
      "====================================================================================\n",
      "\n",
      "\n"
707
     ]
Bannier Delphine's avatar
Bannier Delphine committed
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
    }
   ],
   "source": [
    "from processing.models import fit_and_evaluate\n",
    "n_folds=2\n",
    "epochs=30\n",
    "batch_size=8\n",
    "\n",
    "\n",
    "#save the model history in a list after fitting so that we can plot later\n",
    "model_history = [] \n",
    "\n",
    "for i in range(n_folds):\n",
    "    print(\"Training on Fold: \",i+1)\n",
    "    model = None\n",
    "    model = create_hybrid(trainAttrX.shape[1], shape = (240,240,1))\n",
    "    model.compile(loss=\"mean_absolute_percentage_error\", optimizer=opt)\n",
    "    t_x, val_x, t_y, val_y = custom_shuffle_split(trainAttrX,train_dataset,trainY,test_size = 0.1)    \n",
    "    model_history.append(fit_and_evaluate(t_x, val_x, t_y, val_y, epochs, batch_size,model,es,cp))\n",
    "    print(\"=======\"*12, end=\"\\n\\n\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
735
    {
Bannier Delphine's avatar
Bannier Delphine committed
736
737
738
739
740
741
742
743
744
745
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
Bannier Delphine's avatar
Bannier Delphine committed
746
747
748
    }
   ],
   "source": [
Bannier Delphine's avatar
Bannier Delphine committed
749
750
751
752
753
754
755
    "import matplotlib.pyplot as plt \n",
    "\n",
    "plt.title('Loss vs Epochs')\n",
    "plt.plot(model_history[0].history['loss'], label='Training Fold 1')\n",
    "plt.plot(model_history[1].history['loss'], label='Training Fold 2')\n",
    "plt.legend()\n",
    "plt.show()"
Bannier Delphine's avatar
Bannier Delphine committed
756
757
758
759
   ]
  },
  {
   "cell_type": "code",
Bannier Delphine's avatar
Bannier Delphine committed
760
   "execution_count": 21,
Bannier Delphine's avatar
Bannier Delphine committed
761
762
763
   "metadata": {},
   "outputs": [
    {
Bannier Delphine's avatar
Bannier Delphine committed
764
765
766
767
768
769
770
771
772
773
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
Bannier Delphine's avatar
Bannier Delphine committed
774
775
776
    }
   ],
   "source": [
Bannier Delphine's avatar
Bannier Delphine committed
777
778
779
780
781
782
783
784
785
786
787
788
789
790
    "plt.title('Validation loss vs Epochs')\n",
    "plt.plot(model_history[0].history['val_loss'], label='Training Fold 1')\n",
    "plt.plot(model_history[1].history['val_loss'], label='Training Fold 2')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.models import load_model\n",
Bannier Delphine's avatar
Bannier Delphine committed
791
    "\n",
Bannier Delphine's avatar
Bannier Delphine committed
792
    "model = load_model('superposition_injection.h5')"
Bannier Delphine's avatar
Bannier Delphine committed
793
794
   ]
  },
Bannier Delphine's avatar
Bannier Delphine committed
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
Bannier Delphine's avatar
Bannier Delphine committed
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# F - Evaluation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Training set"
   ]
  },
  {
   "cell_type": "code",
Bannier Delphine's avatar
Bannier Delphine committed
832
   "execution_count": 25,
Bannier Delphine's avatar
Bannier Delphine committed
833
   "metadata": {},
Bannier Delphine's avatar
Bannier Delphine committed
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:predicting ...\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "avg. FVC: 2690.479018721756, std FVC 832.7709592986739\n",
      "mean difference : 27.58%, std: 29.55%\n",
      "competition score : -4.610789567859322\n"
     ]
    }
   ],
Bannier Delphine's avatar
Bannier Delphine committed
852
853
854
   "source": [
    "from postprocessing.evaluate import evaluate_hybrid, compute_score\n",
    "\n",
Bannier Delphine's avatar
Bannier Delphine committed
855
    "preds = evaluate_hybrid(model, df, trainAttrX, train_dataset, trainY, sc)\n",
Bannier Delphine's avatar
Bannier Delphine committed
856
857
858
859
860
861
    "conf, score = compute_score(trainY,preds.flatten())\n",
    "print('competition score :', score)"
   ]
  },
  {
   "cell_type": "code",
Bannier Delphine's avatar
Bannier Delphine committed
862
   "execution_count": 26,
Bannier Delphine's avatar
Bannier Delphine committed
863
   "metadata": {},
Bannier Delphine's avatar
Bannier Delphine committed
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "35/35 [==============================] - 11s 314ms/step - loss: 101.1096\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "101.10960388183594"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
Bannier Delphine's avatar
Bannier Delphine committed
883
   "source": [
Bannier Delphine's avatar
Bannier Delphine committed
884
    "model.evaluate([trainAttrX, train_dataset], trainY)"
Bannier Delphine's avatar
Bannier Delphine committed
885
886
887
888
889
890
891
892
893
894
895
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Test set"
   ]
  },
  {
   "cell_type": "code",
Bannier Delphine's avatar
Bannier Delphine committed
896
   "execution_count": 27,
Bannier Delphine's avatar
Bannier Delphine committed
897
   "metadata": {},
Bannier Delphine's avatar
Bannier Delphine committed
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:predicting ...\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "avg. FVC: 2690.479018721756, std FVC 832.7709592986739\n",
      "mean difference : 34.22%, std: 31.98%\n",
      "competition score : -4.61193964179478\n"
     ]
    }
   ],
Bannier Delphine's avatar
Bannier Delphine committed
916
   "source": [
Bannier Delphine's avatar
Bannier Delphine committed
917
    "preds = evaluate_hybrid(model, df, testAttrX, test_dataset, testY, sc)\n",
Bannier Delphine's avatar
Bannier Delphine committed
918
919
920
921
922
923
    "conf, score = compute_score(testY,preds.flatten())\n",
    "print('competition score :', score)"
   ]
  },
  {
   "cell_type": "code",
Bannier Delphine's avatar
Bannier Delphine committed
924
   "execution_count": 28,
Bannier Delphine's avatar
Bannier Delphine committed
925
   "metadata": {},
Bannier Delphine's avatar
Bannier Delphine committed
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "9/9 [==============================] - 3s 279ms/step - loss: 100.0515\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "100.05147552490234"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
Bannier Delphine's avatar
Bannier Delphine committed
945
   "source": [
Bannier Delphine's avatar
Bannier Delphine committed
946
    "model.evaluate([testAttrX, test_dataset], testY)"
Bannier Delphine's avatar
Bannier Delphine committed
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# G - Sample submission file"
   ]
  },
  {
   "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": "3.8.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}