### 한글 폰트 설정
import matplotlib
from matplotlib import font_manager, rc
import matplotlib.pyplot as plt
import platform
path = "C:/Windows/Fonts/malgun.ttf"
if platform.system() == "Windows":
font_name = font_manager.FontProperties(fname=path).get_name()
rc('font', family=font_name)
elif platform.system()=="Darwin":
rc('font', family='AppleGothic')
else:
print("Unknown System")
import tensorflow as tf
import keras
print(tf.__version__)
print(keras.__version__)
2.9.1 2.9.0
from keras.datasets import imdb
(train_data, train_labels), (test_data, test_labels) = imdb.load_data(num_words=10000)
print(train_data.shape, train_labels.shape)
print(test_data.shape, test_labels.shape)
(25000,) (25000,) (25000,) (25000,)
# train_data의 하나(numpy)에서 10개 정도 확인해 보기
print(type(train_data[0]), len(train_data[0])) # 자료형과 개수
print("하나의 리뷰 단어 개수 : ", len(train_data[0]))
train_data[0][0:15]
<class 'list'> 218 하나의 리뷰 단어 개수 : 218
[1, 14, 22, 16, 43, 530, 973, 1622, 1385, 65, 458, 4468, 66, 3941, 4]
# 5개의 긍정/부정 확인
train_labels[0:5]
array([1, 0, 0, 1, 0], dtype=int64)
[max(sequence) for sequence in train_data][0:10] # 10개 리뷰의 각 리뷰의 단어 인덱스의 최대값
[7486, 9837, 6905, 9941, 7224, 7982, 9363, 9820, 7612, 8419]
max([max(sequence) for sequence in train_data])
9999
# word_index는 단어와 정수 인덱스를 매핑한 딕셔너리입니다
word_index = imdb.get_word_index()
# 전체 단어:인덱스 쌍의 수
print( len(word_index) ) # 88584개
list_word_index = list([ (value, key) for (key, value) in word_index.items() ])
list_word_index[0:10]
88584
[(34701, 'fawn'), (52006, 'tsukino'), (52007, 'nunnery'), (16816, 'sonja'), (63951, 'vani'), (1408, 'woods'), (16115, 'spiders'), (2345, 'hanging'), (2289, 'woody'), (52008, 'trawling')]
# 정수 인덱스와 단어를 매핑하도록 뒤집습니다
reverse_word_index = dict([(value, key) for (key, value) in word_index.items()])
reverse_word_index
{34701: 'fawn',
52006: 'tsukino',
52007: 'nunnery',
16816: 'sonja',
63951: 'vani',
1408: 'woods',
16115: 'spiders',
2345: 'hanging',
2289: 'woody',
52008: 'trawling',
52009: "hold's",
11307: 'comically',
40830: 'localized',
30568: 'disobeying',
52010: "'royale",
40831: "harpo's",
52011: 'canet',
19313: 'aileen',
52012: 'acurately',
52013: "diplomat's",
25242: 'rickman',
6746: 'arranged',
52014: 'rumbustious',
52015: 'familiarness',
52016: "spider'",
68804: 'hahahah',
52017: "wood'",
40833: 'transvestism',
34702: "hangin'",
2338: 'bringing',
40834: 'seamier',
34703: 'wooded',
52018: 'bravora',
16817: 'grueling',
1636: 'wooden',
16818: 'wednesday',
52019: "'prix",
34704: 'altagracia',
52020: 'circuitry',
11585: 'crotch',
57766: 'busybody',
52021: "tart'n'tangy",
14129: 'burgade',
52023: 'thrace',
11038: "tom's",
52025: 'snuggles',
29114: 'francesco',
52027: 'complainers',
52125: 'templarios',
40835: '272',
52028: '273',
52130: 'zaniacs',
34706: '275',
27631: 'consenting',
40836: 'snuggled',
15492: 'inanimate',
52030: 'uality',
11926: 'bronte',
4010: 'errors',
3230: 'dialogs',
52031: "yomada's",
34707: "madman's",
30585: 'dialoge',
52033: 'usenet',
40837: 'videodrome',
26338: "kid'",
52034: 'pawed',
30569: "'girlfriend'",
52035: "'pleasure",
52036: "'reloaded'",
40839: "kazakos'",
52037: 'rocque',
52038: 'mailings',
11927: 'brainwashed',
16819: 'mcanally',
52039: "tom''",
25243: 'kurupt',
21905: 'affiliated',
52040: 'babaganoosh',
40840: "noe's",
40841: 'quart',
359: 'kids',
5034: 'uplifting',
7093: 'controversy',
21906: 'kida',
23379: 'kidd',
52041: "error'",
52042: 'neurologist',
18510: 'spotty',
30570: 'cobblers',
9878: 'projection',
40842: 'fastforwarding',
52043: 'sters',
52044: "eggar's",
52045: 'etherything',
40843: 'gateshead',
34708: 'airball',
25244: 'unsinkable',
7180: 'stern',
52046: "cervi's",
40844: 'dnd',
11586: 'dna',
20598: 'insecurity',
52047: "'reboot'",
11037: 'trelkovsky',
52048: 'jaekel',
52049: 'sidebars',
52050: "sforza's",
17633: 'distortions',
52051: 'mutinies',
30602: 'sermons',
40846: '7ft',
52052: 'boobage',
52053: "o'bannon's",
23380: 'populations',
52054: 'chulak',
27633: 'mesmerize',
52055: 'quinnell',
10307: 'yahoo',
52057: 'meteorologist',
42577: 'beswick',
15493: 'boorman',
40847: 'voicework',
52058: "ster'",
22922: 'blustering',
52059: 'hj',
27634: 'intake',
5621: 'morally',
40849: 'jumbling',
52060: 'bowersock',
52061: "'porky's'",
16821: 'gershon',
40850: 'ludicrosity',
52062: 'coprophilia',
40851: 'expressively',
19500: "india's",
34710: "post's",
52063: 'wana',
5283: 'wang',
30571: 'wand',
25245: 'wane',
52321: 'edgeways',
34711: 'titanium',
40852: 'pinta',
178: 'want',
30572: 'pinto',
52065: 'whoopdedoodles',
21908: 'tchaikovsky',
2103: 'travel',
52066: "'victory'",
11928: 'copious',
22433: 'gouge',
52067: "chapters'",
6702: 'barbra',
30573: 'uselessness',
52068: "wan'",
27635: 'assimilated',
16116: 'petiot',
52069: 'most\x85and',
3930: 'dinosaurs',
352: 'wrong',
52070: 'seda',
52071: 'stollen',
34712: 'sentencing',
40853: 'ouroboros',
40854: 'assimilates',
40855: 'colorfully',
27636: 'glenne',
52072: 'dongen',
4760: 'subplots',
52073: 'kiloton',
23381: 'chandon',
34713: "effect'",
27637: 'snugly',
40856: 'kuei',
9092: 'welcomed',
30071: 'dishonor',
52075: 'concurrence',
23382: 'stoicism',
14896: "guys'",
52077: "beroemd'",
6703: 'butcher',
40857: "melfi's",
30623: 'aargh',
20599: 'playhouse',
11308: 'wickedly',
1180: 'fit',
52078: 'labratory',
40859: 'lifeline',
1927: 'screaming',
4287: 'fix',
52079: 'cineliterate',
52080: 'fic',
52081: 'fia',
34714: 'fig',
52082: 'fmvs',
52083: 'fie',
52084: 'reentered',
30574: 'fin',
52085: 'doctresses',
52086: 'fil',
12606: 'zucker',
31931: 'ached',
52088: 'counsil',
52089: 'paterfamilias',
13885: 'songwriter',
34715: 'shivam',
9654: 'hurting',
299: 'effects',
52090: 'slauther',
52091: "'flame'",
52092: 'sommerset',
52093: 'interwhined',
27638: 'whacking',
52094: 'bartok',
8775: 'barton',
21909: 'frewer',
52095: "fi'",
6192: 'ingrid',
30575: 'stribor',
52096: 'approporiately',
52097: 'wobblyhand',
52098: 'tantalisingly',
52099: 'ankylosaurus',
17634: 'parasites',
52100: 'childen',
52101: "jenkins'",
52102: 'metafiction',
17635: 'golem',
40860: 'indiscretion',
23383: "reeves'",
57781: "inamorata's",
52104: 'brittannica',
7916: 'adapt',
30576: "russo's",
48246: 'guitarists',
10553: 'abbott',
40861: 'abbots',
17649: 'lanisha',
40863: 'magickal',
52105: 'mattter',
52106: "'willy",
34716: 'pumpkins',
52107: 'stuntpeople',
30577: 'estimate',
40864: 'ugghhh',
11309: 'gameplay',
52108: "wern't",
40865: "n'sync",
16117: 'sickeningly',
40866: 'chiara',
4011: 'disturbed',
40867: 'portmanteau',
52109: 'ineffectively',
82143: "duchonvey's",
37519: "nasty'",
1285: 'purpose',
52112: 'lazers',
28105: 'lightened',
52113: 'kaliganj',
52114: 'popularism',
18511: "damme's",
30578: 'stylistics',
52115: 'mindgaming',
46449: 'spoilerish',
52117: "'corny'",
34718: 'boerner',
6792: 'olds',
52118: 'bakelite',
27639: 'renovated',
27640: 'forrester',
52119: "lumiere's",
52024: 'gaskets',
884: 'needed',
34719: 'smight',
1297: 'master',
25905: "edie's",
40868: 'seeber',
52120: 'hiya',
52121: 'fuzziness',
14897: 'genesis',
12607: 'rewards',
30579: 'enthrall',
40869: "'about",
52122: "recollection's",
11039: 'mutilated',
52123: 'fatherlands',
52124: "fischer's",
5399: 'positively',
34705: '270',
34720: 'ahmed',
9836: 'zatoichi',
13886: 'bannister',
52127: 'anniversaries',
30580: "helm's",
52128: "'work'",
34721: 'exclaimed',
52129: "'unfunny'",
52029: '274',
544: 'feeling',
52131: "wanda's",
33266: 'dolan',
52133: '278',
52134: 'peacoat',
40870: 'brawny',
40871: 'mishra',
40872: 'worlders',
52135: 'protags',
52136: 'skullcap',
57596: 'dastagir',
5622: 'affairs',
7799: 'wholesome',
52137: 'hymen',
25246: 'paramedics',
52138: 'unpersons',
52139: 'heavyarms',
52140: 'affaire',
52141: 'coulisses',
40873: 'hymer',
52142: 'kremlin',
30581: 'shipments',
52143: 'pixilated',
30582: "'00s",
18512: 'diminishing',
1357: 'cinematic',
14898: 'resonates',
40874: 'simplify',
40875: "nature'",
40876: 'temptresses',
16822: 'reverence',
19502: 'resonated',
34722: 'dailey',
52144: '2\x85',
27641: 'treize',
52145: 'majo',
21910: 'kiya',
52146: 'woolnough',
39797: 'thanatos',
35731: 'sandoval',
40879: 'dorama',
52147: "o'shaughnessy",
4988: 'tech',
32018: 'fugitives',
30583: 'teck',
76125: "'e'",
40881: 'doesn’t',
52149: 'purged',
657: 'saying',
41095: "martians'",
23418: 'norliss',
27642: 'dickey',
52152: 'dicker',
52153: "'sependipity",
8422: 'padded',
57792: 'ordell',
40882: "sturges'",
52154: 'independentcritics',
5745: 'tempted',
34724: "atkinson's",
25247: 'hounded',
52155: 'apace',
15494: 'clicked',
30584: "'humor'",
17177: "martino's",
52156: "'supporting",
52032: 'warmongering',
34725: "zemeckis's",
21911: 'lube',
52157: 'shocky',
7476: 'plate',
40883: 'plata',
40884: 'sturgess',
40885: "nerds'",
20600: 'plato',
34726: 'plath',
40886: 'platt',
52159: 'mcnab',
27643: 'clumsiness',
3899: 'altogether',
42584: 'massacring',
52160: 'bicenntinial',
40887: 'skaal',
14360: 'droning',
8776: 'lds',
21912: 'jaguar',
34727: "cale's",
1777: 'nicely',
4588: 'mummy',
18513: "lot's",
10086: 'patch',
50202: 'kerkhof',
52161: "leader's",
27644: "'movie",
52162: 'uncomfirmed',
40888: 'heirloom',
47360: 'wrangle',
52163: 'emotion\x85',
52164: "'stargate'",
40889: 'pinoy',
40890: 'conchatta',
41128: 'broeke',
40891: 'advisedly',
17636: "barker's",
52166: 'descours',
772: 'lots',
9259: 'lotr',
9879: 'irs',
52167: 'lott',
40892: 'xvi',
34728: 'irk',
52168: 'irl',
6887: 'ira',
21913: 'belzer',
52169: 'irc',
27645: 'ire',
40893: 'requisites',
7693: 'discipline',
52961: 'lyoko',
11310: 'extend',
873: 'nature',
52170: "'dickie'",
40894: 'optimist',
30586: 'lapping',
3900: 'superficial',
52171: 'vestment',
2823: 'extent',
52172: 'tendons',
52173: "heller's",
52174: 'quagmires',
52175: 'miyako',
20601: 'moocow',
52176: "coles'",
40895: 'lookit',
52177: 'ravenously',
40896: 'levitating',
52178: 'perfunctorily',
30587: 'lookin',
40898: "lot'",
52179: 'lookie',
34870: 'fearlessly',
52181: 'libyan',
40899: 'fondles',
35714: 'gopher',
40901: 'wearying',
52182: "nz's",
27646: 'minuses',
52183: 'puposelessly',
52184: 'shandling',
31268: 'decapitates',
11929: 'humming',
40902: "'nother",
21914: 'smackdown',
30588: 'underdone',
40903: 'frf',
52185: 'triviality',
25248: 'fro',
8777: 'bothers',
52186: "'kensington",
73: 'much',
34730: 'muco',
22615: 'wiseguy',
27648: "richie's",
40904: 'tonino',
52187: 'unleavened',
11587: 'fry',
40905: "'tv'",
40906: 'toning',
14361: 'obese',
30589: 'sensationalized',
40907: 'spiv',
6259: 'spit',
7364: 'arkin',
21915: 'charleton',
16823: 'jeon',
21916: 'boardroom',
4989: 'doubts',
3084: 'spin',
53083: 'hepo',
27649: 'wildcat',
10584: 'venoms',
52191: 'misconstrues',
18514: 'mesmerising',
40908: 'misconstrued',
52192: 'rescinds',
52193: 'prostrate',
40909: 'majid',
16479: 'climbed',
34731: 'canoeing',
52195: 'majin',
57804: 'animie',
40910: 'sylke',
14899: 'conditioned',
40911: 'waddell',
52196: '3\x85',
41188: 'hyperdrive',
34732: 'conditioner',
53153: 'bricklayer',
2576: 'hong',
52198: 'memoriam',
30592: 'inventively',
25249: "levant's",
20638: 'portobello',
52200: 'remand',
19504: 'mummified',
27650: 'honk',
19505: 'spews',
40912: 'visitations',
52201: 'mummifies',
25250: 'cavanaugh',
23385: 'zeon',
40913: "jungle's",
34733: 'viertel',
27651: 'frenchmen',
52202: 'torpedoes',
52203: 'schlessinger',
34734: 'torpedoed',
69876: 'blister',
52204: 'cinefest',
34735: 'furlough',
52205: 'mainsequence',
40914: 'mentors',
9094: 'academic',
20602: 'stillness',
40915: 'academia',
52206: 'lonelier',
52207: 'nibby',
52208: "losers'",
40916: 'cineastes',
4449: 'corporate',
40917: 'massaging',
30593: 'bellow',
19506: 'absurdities',
53241: 'expetations',
40918: 'nyfiken',
75638: 'mehras',
52209: 'lasse',
52210: 'visability',
33946: 'militarily',
52211: "elder'",
19023: 'gainsbourg',
20603: 'hah',
13420: 'hai',
34736: 'haj',
25251: 'hak',
4311: 'hal',
4892: 'ham',
53259: 'duffer',
52213: 'haa',
66: 'had',
11930: 'advancement',
16825: 'hag',
25252: "hand'",
13421: 'hay',
20604: 'mcnamara',
52214: "mozart's",
30731: 'duffel',
30594: 'haq',
13887: 'har',
44: 'has',
2401: 'hat',
40919: 'hav',
30595: 'haw',
52215: 'figtings',
15495: 'elders',
52216: 'underpanted',
52217: 'pninson',
27652: 'unequivocally',
23673: "barbara's",
52219: "bello'",
12997: 'indicative',
40920: 'yawnfest',
52220: 'hexploitation',
52221: "loder's",
27653: 'sleuthing',
32622: "justin's",
52222: "'ball",
52223: "'summer",
34935: "'demons'",
52225: "mormon's",
34737: "laughton's",
52226: 'debell',
39724: 'shipyard',
30597: 'unabashedly',
40401: 'disks',
2290: 'crowd',
10087: 'crowe',
56434: "vancouver's",
34738: 'mosques',
6627: 'crown',
52227: 'culpas',
27654: 'crows',
53344: 'surrell',
52229: 'flowless',
52230: 'sheirk',
40923: "'three",
52231: "peterson'",
52232: 'ooverall',
40924: 'perchance',
1321: 'bottom',
53363: 'chabert',
52233: 'sneha',
13888: 'inhuman',
52234: 'ichii',
52235: 'ursla',
30598: 'completly',
40925: 'moviedom',
52236: 'raddick',
51995: 'brundage',
40926: 'brigades',
1181: 'starring',
52237: "'goal'",
52238: 'caskets',
52239: 'willcock',
52240: "threesome's",
52241: "mosque'",
52242: "cover's",
17637: 'spaceships',
40927: 'anomalous',
27655: 'ptsd',
52243: 'shirdan',
21962: 'obscenity',
30599: 'lemmings',
30600: 'duccio',
52244: "levene's",
52245: "'gorby'",
25255: "teenager's",
5340: 'marshall',
9095: 'honeymoon',
3231: 'shoots',
12258: 'despised',
52246: 'okabasho',
8289: 'fabric',
18515: 'cannavale',
3537: 'raped',
52247: "tutt's",
17638: 'grasping',
18516: 'despises',
40928: "thief's",
8926: 'rapes',
52248: 'raper',
27656: "eyre'",
52249: 'walchek',
23386: "elmo's",
40929: 'perfumes',
21918: 'spurting',
52250: "exposition'\x85",
52251: 'denoting',
34740: 'thesaurus',
40930: "shoot'",
49759: 'bonejack',
52253: 'simpsonian',
30601: 'hebetude',
34741: "hallow's",
52254: 'desperation\x85',
34742: 'incinerator',
10308: 'congratulations',
52255: 'humbled',
5924: "else's",
40845: 'trelkovski',
52256: "rape'",
59386: "'chapters'",
52257: '1600s',
7253: 'martian',
25256: 'nicest',
52259: 'eyred',
9457: 'passenger',
6041: 'disgrace',
52260: 'moderne',
5120: 'barrymore',
52261: 'yankovich',
40931: 'moderns',
52262: 'studliest',
52263: 'bedsheet',
14900: 'decapitation',
52264: 'slurring',
52265: "'nunsploitation'",
34743: "'character'",
9880: 'cambodia',
52266: 'rebelious',
27657: 'pasadena',
40932: 'crowne',
52267: "'bedchamber",
52268: 'conjectural',
52269: 'appologize',
52270: 'halfassing',
57816: 'paycheque',
20606: 'palms',
52271: "'islands",
40933: 'hawked',
21919: 'palme',
40934: 'conservatively',
64007: 'larp',
5558: 'palma',
21920: 'smelling',
12998: 'aragorn',
52272: 'hawker',
52273: 'hawkes',
3975: 'explosions',
8059: 'loren',
52274: "pyle's",
6704: 'shootout',
18517: "mike's",
52275: "driscoll's",
40935: 'cogsworth',
52276: "britian's",
34744: 'childs',
52277: "portrait's",
3626: 'chain',
2497: 'whoever',
52278: 'puttered',
52279: 'childe',
52280: 'maywether',
3036: 'chair',
52281: "rance's",
34745: 'machu',
4517: 'ballet',
34746: 'grapples',
76152: 'summerize',
30603: 'freelance',
52283: "andrea's",
52284: '\x91very',
45879: 'coolidge',
18518: 'mache',
52285: 'balled',
40937: 'grappled',
18519: 'macha',
21921: 'underlining',
5623: 'macho',
19507: 'oversight',
25257: 'machi',
11311: 'verbally',
21922: 'tenacious',
40938: 'windshields',
18557: 'paychecks',
3396: 'jerk',
11931: "good'",
34748: 'prancer',
21923: 'prances',
52286: 'olympus',
21924: 'lark',
10785: 'embark',
7365: 'gloomy',
52287: 'jehaan',
52288: 'turaqui',
20607: "child'",
2894: 'locked',
52289: 'pranced',
2588: 'exact',
52290: 'unattuned',
783: 'minute',
16118: 'skewed',
40940: 'hodgins',
34749: 'skewer',
52291: 'think\x85',
38765: 'rosenstein',
52292: 'helmit',
34750: 'wrestlemanias',
16826: 'hindered',
30604: "martha's",
52293: 'cheree',
52294: "pluckin'",
40941: 'ogles',
11932: 'heavyweight',
82190: 'aada',
11312: 'chopping',
61534: 'strongboy',
41342: 'hegemonic',
40942: 'adorns',
41346: 'xxth',
34751: 'nobuhiro',
52298: 'capitães',
52299: 'kavogianni',
13422: 'antwerp',
6538: 'celebrated',
52300: 'roarke',
40943: 'baggins',
31270: 'cheeseburgers',
52301: 'matras',
52302: "nineties'",
52303: "'craig'",
12999: 'celebrates',
3383: 'unintentionally',
14362: 'drafted',
52304: 'climby',
52305: '303',
18520: 'oldies',
9096: 'climbs',
9655: 'honour',
34752: 'plucking',
30074: '305',
5514: 'address',
40944: 'menjou',
42592: "'freak'",
19508: 'dwindling',
9458: 'benson',
52307: 'white’s',
40945: 'shamelessness',
21925: 'impacted',
52308: 'upatz',
3840: 'cusack',
37567: "flavia's",
52309: 'effette',
34753: 'influx',
52310: 'boooooooo',
52311: 'dimitrova',
13423: 'houseman',
25259: 'bigas',
52312: 'boylen',
52313: 'phillipenes',
40946: 'fakery',
27658: "grandpa's",
27659: 'darnell',
19509: 'undergone',
52315: 'handbags',
21926: 'perished',
37778: 'pooped',
27660: 'vigour',
3627: 'opposed',
52316: 'etude',
11799: "caine's",
52317: 'doozers',
34754: 'photojournals',
52318: 'perishes',
34755: 'constrains',
40948: 'migenes',
30605: 'consoled',
16827: 'alastair',
52319: 'wvs',
52320: 'ooooooh',
34756: 'approving',
40949: 'consoles',
52064: 'disparagement',
52322: 'futureistic',
52323: 'rebounding',
52324: "'date",
52325: 'gregoire',
21927: 'rutherford',
34757: 'americanised',
82196: 'novikov',
1042: 'following',
34758: 'munroe',
52326: "morita'",
52327: 'christenssen',
23106: 'oatmeal',
25260: 'fossey',
40950: 'livered',
13000: 'listens',
76164: "'marci",
52330: "otis's",
23387: 'thanking',
16019: 'maude',
34759: 'extensions',
52332: 'ameteurish',
52333: "commender's",
27661: 'agricultural',
4518: 'convincingly',
17639: 'fueled',
54014: 'mahattan',
40952: "paris's",
52336: 'vulkan',
52337: 'stapes',
52338: 'odysessy',
12259: 'harmon',
4252: 'surfing',
23494: 'halloran',
49580: 'unbelieveably',
52339: "'offed'",
30607: 'quadrant',
19510: 'inhabiting',
34760: 'nebbish',
40953: 'forebears',
34761: 'skirmish',
52340: 'ocassionally',
52341: "'resist",
21928: 'impactful',
52342: 'spicier',
40954: 'touristy',
52343: "'football'",
40955: 'webpage',
52345: 'exurbia',
52346: 'jucier',
14901: 'professors',
34762: 'structuring',
30608: 'jig',
40956: 'overlord',
25261: 'disconnect',
82201: 'sniffle',
40957: 'slimeball',
40958: 'jia',
16828: 'milked',
40959: 'banjoes',
1237: 'jim',
52348: 'workforces',
52349: 'jip',
52350: 'rotweiller',
34763: 'mundaneness',
52351: "'ninja'",
11040: "dead'",
40960: "cipriani's",
20608: 'modestly',
52352: "professor'",
40961: 'shacked',
34764: 'bashful',
23388: 'sorter',
16120: 'overpowering',
18521: 'workmanlike',
27662: 'henpecked',
18522: 'sorted',
52354: "jōb's",
52355: "'always",
34765: "'baptists",
52356: 'dreamcatchers',
52357: "'silence'",
21929: 'hickory',
52358: 'fun\x97yet',
52359: 'breakumentary',
15496: 'didn',
52360: 'didi',
52361: 'pealing',
40962: 'dispite',
25262: "italy's",
21930: 'instability',
6539: 'quarter',
12608: 'quartet',
52362: 'padmé',
52363: "'bleedmedry",
52364: 'pahalniuk',
52365: 'honduras',
10786: 'bursting',
41465: "pablo's",
52367: 'irremediably',
40963: 'presages',
57832: 'bowlegged',
65183: 'dalip',
6260: 'entering',
76172: 'newsradio',
54150: 'presaged',
27663: "giallo's",
40964: 'bouyant',
52368: 'amerterish',
18523: 'rajni',
30610: 'leeves',
34767: 'macauley',
612: 'seriously',
52369: 'sugercoma',
52370: 'grimstead',
52371: "'fairy'",
30611: 'zenda',
52372: "'twins'",
17640: 'realisation',
27664: 'highsmith',
7817: 'raunchy',
40965: 'incentives',
52374: 'flatson',
35097: 'snooker',
16829: 'crazies',
14902: 'crazier',
7094: 'grandma',
52375: 'napunsaktha',
30612: 'workmanship',
52376: 'reisner',
61306: "sanford's",
52377: '\x91doña',
6108: 'modest',
19153: "everything's",
40966: 'hamer',
52379: "couldn't'",
13001: 'quibble',
52380: 'socking',
21931: 'tingler',
52381: 'gutman',
40967: 'lachlan',
52382: 'tableaus',
52383: 'headbanger',
2847: 'spoken',
34768: 'cerebrally',
23490: "'road",
21932: 'tableaux',
40968: "proust's",
40969: 'periodical',
52385: "shoveller's",
25263: 'tamara',
17641: 'affords',
3249: 'concert',
87955: "yara's",
52386: 'someome',
8424: 'lingering',
41511: "abraham's",
34769: 'beesley',
34770: 'cherbourg',
28624: 'kagan',
9097: 'snatch',
9260: "miyazaki's",
25264: 'absorbs',
40970: "koltai's",
64027: 'tingled',
19511: 'crossroads',
16121: 'rehab',
52389: 'falworth',
52390: 'sequals',
...}
print(type(reverse_word_index))
dir(reverse_word_index)[-11:] # 딕셔너리 기능 확인
<class 'dict'>
['clear', 'copy', 'fromkeys', 'get', 'items', 'keys', 'pop', 'popitem', 'setdefault', 'update', 'values']
# reverse_word_index.get(인덱스) # 인덱스에 해당되는 단어가 출력
# reverse_word_index.get(인덱스, '?') # 인덱스에 해당되는 단어가 출력되는데, 단어가 없으면 ? 출력
for i in range(0, 50, 1):
print( reverse_word_index.get(i, '?'), end= " " )
print("train 데이터 첫번째 3단어 살펴보기")
print("index 14 : ", reverse_word_index.get(14-3, '?'))
print("index 22 : ", reverse_word_index.get(22-3, '?'))
print("index 16 : ", reverse_word_index.get(16-3, '?'))
? the and a of to is br in it i this that was as for with movie but film on not you are his have he be one all at by an they who so from like her or just about it's out has if some there what good train 데이터 첫번째 3단어 살펴보기 index 14 : this index 22 : film index 16 : was
print( train_data[0] ) # 첫번째 리뷰(숫자 인덱스)
[1, 14, 22, 16, 43, 530, 973, 1622, 1385, 65, 458, 4468, 66, 3941, 4, 173, 36, 256, 5, 25, 100, 43, 838, 112, 50, 670, 2, 9, 35, 480, 284, 5, 150, 4, 172, 112, 167, 2, 336, 385, 39, 4, 172, 4536, 1111, 17, 546, 38, 13, 447, 4, 192, 50, 16, 6, 147, 2025, 19, 14, 22, 4, 1920, 4613, 469, 4, 22, 71, 87, 12, 16, 43, 530, 38, 76, 15, 13, 1247, 4, 22, 17, 515, 17, 12, 16, 626, 18, 2, 5, 62, 386, 12, 8, 316, 8, 106, 5, 4, 2223, 5244, 16, 480, 66, 3785, 33, 4, 130, 12, 16, 38, 619, 5, 25, 124, 51, 36, 135, 48, 25, 1415, 33, 6, 22, 12, 215, 28, 77, 52, 5, 14, 407, 16, 82, 2, 8, 4, 107, 117, 5952, 15, 256, 4, 2, 7, 3766, 5, 723, 36, 71, 43, 530, 476, 26, 400, 317, 46, 7, 4, 2, 1029, 13, 104, 88, 4, 381, 15, 297, 98, 32, 2071, 56, 26, 141, 6, 194, 7486, 18, 4, 226, 22, 21, 134, 476, 26, 480, 5, 144, 30, 5535, 18, 51, 36, 28, 224, 92, 25, 104, 4, 226, 65, 16, 38, 1334, 88, 12, 16, 283, 5, 16, 4472, 113, 103, 32, 15, 16, 5345, 19, 178, 32]
# train_data[0] : 하나의 리뷰(3, 6, 2, 5, 10...) => 218단어 ...
# [reverse_word_index.get(i - 3, '?') for i in train_data[0]]
print(len( train_data[0]) ) # 첫번째 리뷰는 218개 인덱스(단어)로 이루어져 있다.
print([i for i in train_data[0]]) # train_data[0]인덱스가 for문이 돌아가면서 리스트 형태로 만들어진다.
218 [1, 14, 22, 16, 43, 530, 973, 1622, 1385, 65, 458, 4468, 66, 3941, 4, 173, 36, 256, 5, 25, 100, 43, 838, 112, 50, 670, 2, 9, 35, 480, 284, 5, 150, 4, 172, 112, 167, 2, 336, 385, 39, 4, 172, 4536, 1111, 17, 546, 38, 13, 447, 4, 192, 50, 16, 6, 147, 2025, 19, 14, 22, 4, 1920, 4613, 469, 4, 22, 71, 87, 12, 16, 43, 530, 38, 76, 15, 13, 1247, 4, 22, 17, 515, 17, 12, 16, 626, 18, 2, 5, 62, 386, 12, 8, 316, 8, 106, 5, 4, 2223, 5244, 16, 480, 66, 3785, 33, 4, 130, 12, 16, 38, 619, 5, 25, 124, 51, 36, 135, 48, 25, 1415, 33, 6, 22, 12, 215, 28, 77, 52, 5, 14, 407, 16, 82, 2, 8, 4, 107, 117, 5952, 15, 256, 4, 2, 7, 3766, 5, 723, 36, 71, 43, 530, 476, 26, 400, 317, 46, 7, 4, 2, 1029, 13, 104, 88, 4, 381, 15, 297, 98, 32, 2071, 56, 26, 141, 6, 194, 7486, 18, 4, 226, 22, 21, 134, 476, 26, 480, 5, 144, 30, 5535, 18, 51, 36, 28, 224, 92, 25, 104, 4, 226, 65, 16, 38, 1334, 88, 12, 16, 283, 5, 16, 4472, 113, 103, 32, 15, 16, 5345, 19, 178, 32]
# 두번째 리뷰 - 189개 인덱스(단어)로 이루어져 있고, 이에 대한 리스트를 만들어 출력해 본다.
print(len( train_data[1] )) # 첫번째 리뷰는 218개 인덱스(단어)로 이루어져 있다.
print([i for i in train_data[1] ]) # train_data[0]인덱스가 for문이 돌아가면서 리스트 형태로 만들어진다.
189 [1, 194, 1153, 194, 8255, 78, 228, 5, 6, 1463, 4369, 5012, 134, 26, 4, 715, 8, 118, 1634, 14, 394, 20, 13, 119, 954, 189, 102, 5, 207, 110, 3103, 21, 14, 69, 188, 8, 30, 23, 7, 4, 249, 126, 93, 4, 114, 9, 2300, 1523, 5, 647, 4, 116, 9, 35, 8163, 4, 229, 9, 340, 1322, 4, 118, 9, 4, 130, 4901, 19, 4, 1002, 5, 89, 29, 952, 46, 37, 4, 455, 9, 45, 43, 38, 1543, 1905, 398, 4, 1649, 26, 6853, 5, 163, 11, 3215, 2, 4, 1153, 9, 194, 775, 7, 8255, 2, 349, 2637, 148, 605, 2, 8003, 15, 123, 125, 68, 2, 6853, 15, 349, 165, 4362, 98, 5, 4, 228, 9, 43, 2, 1157, 15, 299, 120, 5, 120, 174, 11, 220, 175, 136, 50, 9, 4373, 228, 8255, 5, 2, 656, 245, 2350, 5, 4, 9837, 131, 152, 491, 18, 2, 32, 7464, 1212, 14, 9, 6, 371, 78, 22, 625, 64, 1382, 9, 8, 168, 145, 23, 4, 1690, 15, 16, 4, 1355, 5, 28, 6, 52, 154, 462, 33, 89, 78, 285, 16, 145, 95]
# [reverse_word_index.get(i - 3, '?') for i in train_data[0]]
# 의미 : 각각의 인덱스(218)에 대한 단어를 리스트 형태로 만든다.
print("첫번째 리뷰의 인덱스를 단어로 매칭시켜서 보여준 것")
print([reverse_word_index.get(i - 3, '?') for i in train_data[0]])
# ' '.join([reverse_word_index.get(i - 3, '?') for i in train_data[0]])
# 각각의 단어 리스트를 공백하나를 넣어주면서 하나의 문자열로 묶어 준것.
첫번째 리뷰의 인덱스를 단어로 매칭시켜서 보여준 것 ['?', 'this', 'film', 'was', 'just', 'brilliant', 'casting', 'location', 'scenery', 'story', 'direction', "everyone's", 'really', 'suited', 'the', 'part', 'they', 'played', 'and', 'you', 'could', 'just', 'imagine', 'being', 'there', 'robert', '?', 'is', 'an', 'amazing', 'actor', 'and', 'now', 'the', 'same', 'being', 'director', '?', 'father', 'came', 'from', 'the', 'same', 'scottish', 'island', 'as', 'myself', 'so', 'i', 'loved', 'the', 'fact', 'there', 'was', 'a', 'real', 'connection', 'with', 'this', 'film', 'the', 'witty', 'remarks', 'throughout', 'the', 'film', 'were', 'great', 'it', 'was', 'just', 'brilliant', 'so', 'much', 'that', 'i', 'bought', 'the', 'film', 'as', 'soon', 'as', 'it', 'was', 'released', 'for', '?', 'and', 'would', 'recommend', 'it', 'to', 'everyone', 'to', 'watch', 'and', 'the', 'fly', 'fishing', 'was', 'amazing', 'really', 'cried', 'at', 'the', 'end', 'it', 'was', 'so', 'sad', 'and', 'you', 'know', 'what', 'they', 'say', 'if', 'you', 'cry', 'at', 'a', 'film', 'it', 'must', 'have', 'been', 'good', 'and', 'this', 'definitely', 'was', 'also', '?', 'to', 'the', 'two', 'little', "boy's", 'that', 'played', 'the', '?', 'of', 'norman', 'and', 'paul', 'they', 'were', 'just', 'brilliant', 'children', 'are', 'often', 'left', 'out', 'of', 'the', '?', 'list', 'i', 'think', 'because', 'the', 'stars', 'that', 'play', 'them', 'all', 'grown', 'up', 'are', 'such', 'a', 'big', 'profile', 'for', 'the', 'whole', 'film', 'but', 'these', 'children', 'are', 'amazing', 'and', 'should', 'be', 'praised', 'for', 'what', 'they', 'have', 'done', "don't", 'you', 'think', 'the', 'whole', 'story', 'was', 'so', 'lovely', 'because', 'it', 'was', 'true', 'and', 'was', "someone's", 'life', 'after', 'all', 'that', 'was', 'shared', 'with', 'us', 'all']
# 0, 1, 2는 '패딩', '문서 시작', '사전에 없음'을 위한 인덱스이므로 3을 뺍니다
decoded_review = ' '.join([reverse_word_index.get(i - 3, '?') for i in train_data[0]])
decoded_review
"? this film was just brilliant casting location scenery story direction everyone's really suited the part they played and you could just imagine being there robert ? is an amazing actor and now the same being director ? father came from the same scottish island as myself so i loved the fact there was a real connection with this film the witty remarks throughout the film were great it was just brilliant so much that i bought the film as soon as it was released for ? and would recommend it to everyone to watch and the fly fishing was amazing really cried at the end it was so sad and you know what they say if you cry at a film it must have been good and this definitely was also ? to the two little boy's that played the ? of norman and paul they were just brilliant children are often left out of the ? list i think because the stars that play them all grown up are such a big profile for the whole film but these children are amazing and should be praised for what they have done don't you think the whole story was so lovely because it was true and was someone's life after all that was shared with us all"
원핫 인코딩
나의 첫번째 리뷰
나의 두번째 리뷰
나의 세번째 리뷰
또 다른 표현
나의 세번째 리뷰
import numpy as np
def vectorize_sequences(sequences, dimension=10000):
# 크기가 (len(sequences), dimension))이고 모든 원소가 0인 행렬을 만듭니다
results = np.zeros((len(sequences), dimension))
for i, sequence in enumerate(sequences):
results[i, sequence] = 1. # results[i]에서 특정 인덱스의 위치를 1로 만듭니다
return results
print("변환 전 : ", train_data.shape)
X_train = vectorize_sequences(train_data)
print("변환 후 : ", X_train.shape)
X_train
변환 전 : (25000,) 변환 후 : (25000, 10000)
array([[0., 1., 1., ..., 0., 0., 0.],
[0., 1., 1., ..., 0., 0., 0.],
[0., 1., 1., ..., 0., 0., 0.],
...,
[0., 1., 1., ..., 0., 0., 0.],
[0., 1., 1., ..., 0., 0., 0.],
[0., 1., 1., ..., 0., 0., 0.]])
print("변환 전 : ",test_data.shape)
X_test = vectorize_sequences(test_data)
print("변환 후 : ",X_test.shape)
X_test
변환 전 : (25000,) 변환 후 : (25000, 10000)
array([[0., 1., 1., ..., 0., 0., 0.],
[0., 1., 1., ..., 0., 0., 0.],
[0., 1., 1., ..., 0., 0., 0.],
...,
[0., 1., 1., ..., 0., 0., 0.],
[0., 1., 1., ..., 0., 0., 0.],
[0., 1., 1., ..., 0., 0., 0.]])
a = [1,2]
np.asarray(a)
array([1, 2])
print(type(train_labels), type(test_labels))
print(train_labels.shape, test_labels.shape)
<class 'numpy.ndarray'> <class 'numpy.ndarray'> (25000,) (25000,)
y_train = np.asarray(train_labels).astype('float32')
y_test = np.asarray(test_labels).astype('float32')
print(type(y_train), type(y_test))
y_train.shape, y_test.shape
<class 'numpy.ndarray'> <class 'numpy.ndarray'>
((25000,), (25000,))
y_train
array([1., 0., 0., ..., 0., 1., 0.], dtype=float32)
y_test
array([0., 1., 1., ..., 0., 0., 0.], dtype=float32)
from keras import models
from keras import layers
model = models.Sequential()
model.add(layers.Dense(16, activation='relu', input_shape=(10000,)))
model.add(layers.Dense(16, activation='relu'))
model.add(layers.Dense(1, activation='sigmoid'))
model.compile(optimizer='rmsprop',
loss='binary_crossentropy',
metrics=['accuracy'])
from tensorflow.keras import optimizers
model.compile(optimizer=optimizers.RMSprop(learning_rate=0.01),
loss='binary_crossentropy',
metrics=['accuracy'] )
from keras import losses
from keras import metrics
model.compile(optimizer=optimizers.RMSprop(lr=0.001),
loss=losses.binary_crossentropy,
metrics=[metrics.binary_accuracy])X_train.shape, y_train.shape
((25000, 10000), (25000,))
X_val = X_train[:10000] # 자체 검증
partial_X_train = X_train[10000:] # 학습용 10000~25000
y_val = y_train[:10000] # 자체 검증
partial_y_train = y_train[10000:] # 학습용 10000~25000
history = model.fit(partial_X_train,
partial_y_train,
epochs=20,
batch_size=512,
validation_data=(X_val, y_val))
Epoch 1/20 30/30 [==============================] - 2s 33ms/step - loss: 0.6185 - accuracy: 0.7185 - val_loss: 0.5058 - val_accuracy: 0.7568 Epoch 2/20 30/30 [==============================] - 1s 22ms/step - loss: 0.3194 - accuracy: 0.8754 - val_loss: 0.3907 - val_accuracy: 0.8449 Epoch 3/20 30/30 [==============================] - 1s 22ms/step - loss: 0.2129 - accuracy: 0.9161 - val_loss: 0.4216 - val_accuracy: 0.8510 Epoch 4/20 30/30 [==============================] - 1s 22ms/step - loss: 0.1684 - accuracy: 0.9364 - val_loss: 0.4502 - val_accuracy: 0.8439 Epoch 5/20 30/30 [==============================] - 1s 22ms/step - loss: 0.1206 - accuracy: 0.9571 - val_loss: 0.3564 - val_accuracy: 0.8818 Epoch 6/20 30/30 [==============================] - 1s 22ms/step - loss: 0.0937 - accuracy: 0.9696 - val_loss: 0.5479 - val_accuracy: 0.8514 Epoch 7/20 30/30 [==============================] - 1s 22ms/step - loss: 0.1148 - accuracy: 0.9612 - val_loss: 0.3717 - val_accuracy: 0.8778 Epoch 8/20 30/30 [==============================] - 1s 23ms/step - loss: 0.0657 - accuracy: 0.9793 - val_loss: 0.4201 - val_accuracy: 0.8771 Epoch 9/20 30/30 [==============================] - 1s 23ms/step - loss: 0.0197 - accuracy: 0.9941 - val_loss: 0.6530 - val_accuracy: 0.8678 Epoch 10/20 30/30 [==============================] - 1s 22ms/step - loss: 0.0907 - accuracy: 0.9826 - val_loss: 0.6377 - val_accuracy: 0.8759 Epoch 11/20 30/30 [==============================] - 1s 23ms/step - loss: 0.0054 - accuracy: 0.9993 - val_loss: 0.7571 - val_accuracy: 0.8730 Epoch 12/20 30/30 [==============================] - 1s 22ms/step - loss: 0.0912 - accuracy: 0.9839 - val_loss: 0.7324 - val_accuracy: 0.8717 Epoch 13/20 30/30 [==============================] - 1s 22ms/step - loss: 0.0027 - accuracy: 0.9998 - val_loss: 0.8392 - val_accuracy: 0.8735 Epoch 14/20 30/30 [==============================] - 1s 22ms/step - loss: 0.0015 - accuracy: 0.9997 - val_loss: 0.9948 - val_accuracy: 0.8704 Epoch 15/20 30/30 [==============================] - 1s 22ms/step - loss: 0.1047 - accuracy: 0.9872 - val_loss: 0.9747 - val_accuracy: 0.8626 Epoch 16/20 30/30 [==============================] - 1s 22ms/step - loss: 0.0015 - accuracy: 0.9999 - val_loss: 1.0148 - val_accuracy: 0.8699 Epoch 17/20 30/30 [==============================] - 1s 22ms/step - loss: 4.2119e-04 - accuracy: 0.9999 - val_loss: 1.2154 - val_accuracy: 0.8697 Epoch 18/20 30/30 [==============================] - 1s 23ms/step - loss: 2.1195e-04 - accuracy: 0.9999 - val_loss: 1.3966 - val_accuracy: 0.8689 Epoch 19/20 30/30 [==============================] - 1s 22ms/step - loss: 1.5039e-04 - accuracy: 0.9999 - val_loss: 1.5292 - val_accuracy: 0.8687 Epoch 20/20 30/30 [==============================] - 1s 20ms/step - loss: 3.7695e-06 - accuracy: 1.0000 - val_loss: 1.6868 - val_accuracy: 0.8699
history_dict = history.history
history_dict.keys()
dict_keys(['loss', 'accuracy', 'val_loss', 'val_accuracy'])
import matplotlib.pyplot as plt
acc = history.history['accuracy']
val_acc = history.history['val_accuracy']
loss = history.history['loss']
val_loss = history.history['val_loss']
epochs = range(1, len(acc) + 1)
plt.figure(figsize=(12,7))
plt.subplot(1,2,1)
# ‘bo’는 파란색 점을 의미합니다
plt.plot(epochs, loss, 'bo', label='Training loss')
# ‘b’는 파란색 실선을 의미합니다
plt.plot(epochs, val_loss, 'b', label='Validation loss')
plt.title('학습과 평가 데이터 셋 - 손실')
plt.xlabel('Epochs')
plt.ylabel('Loss')
plt.legend()
plt.subplot(1,2,2)
acc = history_dict['accuracy']
val_acc = history_dict['val_accuracy']
plt.plot(epochs, acc, 'bo', label='Training acc')
plt.plot(epochs, val_acc, 'b', label='Validation acc')
plt.title('학습과 평가 데이터 셋 - 정확도')
plt.xlabel('Epochs')
plt.ylabel('Accuracy')
plt.legend()
plt.show()
model = models.Sequential()
model.add(layers.Dense(16, activation='relu', input_shape=(10000,)))
model.add(layers.Dense(16, activation='relu'))
model.add(layers.Dense(1, activation='sigmoid'))
model.compile(optimizer='rmsprop',
loss='binary_crossentropy',
metrics=['accuracy'])
hist = model.fit(X_train, y_train, epochs=4, batch_size=512, validation_data=(X_val, y_val))
results = model.evaluate(X_test, y_test)
Epoch 1/4 49/49 [==============================] - 2s 24ms/step - loss: 0.4827 - accuracy: 0.8156 - val_loss: 0.3221 - val_accuracy: 0.9048 Epoch 2/4 49/49 [==============================] - 1s 18ms/step - loss: 0.2778 - accuracy: 0.9088 - val_loss: 0.2105 - val_accuracy: 0.9367 Epoch 3/4 49/49 [==============================] - 1s 19ms/step - loss: 0.2094 - accuracy: 0.9268 - val_loss: 0.1765 - val_accuracy: 0.9422 Epoch 4/4 49/49 [==============================] - 1s 18ms/step - loss: 0.1763 - accuracy: 0.9392 - val_loss: 0.1520 - val_accuracy: 0.9503 782/782 [==============================] - 2s 2ms/step - loss: 0.2963 - accuracy: 0.8814
results
[0.2962694466114044, 0.8813999891281128]
history_dict = hist.history
history_dict.keys()
dict_keys(['loss', 'accuracy', 'val_loss', 'val_accuracy'])
acc = hist.history['accuracy']
val_acc = hist.history['val_accuracy']
loss = hist.history['loss']
val_loss = hist.history['val_loss']
epochs = range(1, len(acc) + 1)
plt.figure(figsize=(12,7))
plt.subplot(1,2,1)
# ‘bo’는 파란색 점을 의미합니다
plt.plot(epochs, loss, 'bo', label='Training loss')
# ‘b’는 파란색 실선을 의미합니다
plt.plot(epochs, val_loss, 'b', label='Validation loss')
plt.title('학습과 평가 데이터 셋 - 손실')
plt.xlabel('Epochs')
plt.ylabel('Loss')
plt.legend()
plt.subplot(1,2,2)
acc = history_dict['accuracy']
val_acc = history_dict['val_accuracy']
plt.plot(epochs, acc, 'bo', label='Training acc')
plt.plot(epochs, val_acc, 'b', label='Validation acc')
plt.title('학습과 평가 데이터 셋 - 정확도')
plt.xlabel('Epochs')
plt.ylabel('Accuracy')
plt.legend()
plt.show()
model.predict(X_test)
782/782 [==============================] - 2s 2ms/step
array([[0.2491391 ],
[0.9993358 ],
[0.92614776],
...,
[0.15147226],
[0.11113814],
[0.67956156]], dtype=float32)
model = models.Sequential()
model.add(layers.Dense(16, activation='relu', input_shape=(10000,)))
model.add(layers.Dense(16, activation='relu'))
model.add(layers.Dense(1, activation='sigmoid'))
model.compile(optimizer=optimizers.RMSprop(learning_rate=0.001),
loss='binary_crossentropy',
metrics=['accuracy'])
hist2 = model.fit(partial_X_train,
partial_y_train,
epochs=20,
batch_size=128,
validation_data=(X_val, y_val))
Epoch 1/20 118/118 [==============================] - 2s 11ms/step - loss: 0.4115 - accuracy: 0.8371 - val_loss: 0.2875 - val_accuracy: 0.8887 Epoch 2/20 118/118 [==============================] - 1s 9ms/step - loss: 0.2118 - accuracy: 0.9223 - val_loss: 0.2787 - val_accuracy: 0.8876 Epoch 3/20 118/118 [==============================] - 1s 9ms/step - loss: 0.1554 - accuracy: 0.9413 - val_loss: 0.3085 - val_accuracy: 0.8834 Epoch 4/20 118/118 [==============================] - 1s 9ms/step - loss: 0.1187 - accuracy: 0.9581 - val_loss: 0.3329 - val_accuracy: 0.8812 Epoch 5/20 118/118 [==============================] - 1s 9ms/step - loss: 0.0897 - accuracy: 0.9687 - val_loss: 0.3767 - val_accuracy: 0.8758 Epoch 6/20 118/118 [==============================] - 1s 9ms/step - loss: 0.0675 - accuracy: 0.9771 - val_loss: 0.4446 - val_accuracy: 0.8713 Epoch 7/20 118/118 [==============================] - 1s 9ms/step - loss: 0.0480 - accuracy: 0.9843 - val_loss: 0.5111 - val_accuracy: 0.8693 Epoch 8/20 118/118 [==============================] - 1s 9ms/step - loss: 0.0339 - accuracy: 0.9887 - val_loss: 0.5585 - val_accuracy: 0.8676 Epoch 9/20 118/118 [==============================] - 1s 9ms/step - loss: 0.0233 - accuracy: 0.9923 - val_loss: 0.6355 - val_accuracy: 0.8644 Epoch 10/20 118/118 [==============================] - 1s 9ms/step - loss: 0.0147 - accuracy: 0.9957 - val_loss: 0.7187 - val_accuracy: 0.8601 Epoch 11/20 118/118 [==============================] - 1s 9ms/step - loss: 0.0099 - accuracy: 0.9977 - val_loss: 0.8024 - val_accuracy: 0.8609 Epoch 12/20 118/118 [==============================] - 1s 9ms/step - loss: 0.0052 - accuracy: 0.9989 - val_loss: 0.9183 - val_accuracy: 0.8602 Epoch 13/20 118/118 [==============================] - 1s 9ms/step - loss: 0.0029 - accuracy: 0.9995 - val_loss: 1.0095 - val_accuracy: 0.8553 Epoch 14/20 118/118 [==============================] - 1s 9ms/step - loss: 0.0018 - accuracy: 0.9996 - val_loss: 1.1316 - val_accuracy: 0.8581 Epoch 15/20 118/118 [==============================] - 1s 9ms/step - loss: 8.1083e-04 - accuracy: 0.9999 - val_loss: 1.2461 - val_accuracy: 0.8585 Epoch 16/20 118/118 [==============================] - 1s 9ms/step - loss: 3.3346e-04 - accuracy: 1.0000 - val_loss: 1.3680 - val_accuracy: 0.8563 Epoch 17/20 118/118 [==============================] - 1s 9ms/step - loss: 4.4051e-04 - accuracy: 0.9999 - val_loss: 1.5174 - val_accuracy: 0.8544 Epoch 18/20 118/118 [==============================] - 1s 9ms/step - loss: 5.0479e-05 - accuracy: 1.0000 - val_loss: 1.6037 - val_accuracy: 0.8543 Epoch 19/20 118/118 [==============================] - 1s 9ms/step - loss: 1.6217e-04 - accuracy: 1.0000 - val_loss: 1.7044 - val_accuracy: 0.8563 Epoch 20/20 118/118 [==============================] - 1s 9ms/step - loss: 1.1381e-04 - accuracy: 0.9999 - val_loss: 1.7591 - val_accuracy: 0.8558
history_dict = hist2.history
history_dict.keys()
dict_keys(['loss', 'accuracy', 'val_loss', 'val_accuracy'])
import matplotlib.pyplot as plt
acc = hist2.history['accuracy']
val_acc = hist2.history['val_accuracy']
loss = hist2.history['loss']
val_loss = hist2.history['val_loss']
epochs = range(1, len(acc) + 1)
plt.figure(figsize=(12,7))
plt.subplot(1,2,1)
# ‘bo’는 파란색 점을 의미합니다
plt.plot(epochs, loss, 'bo', label='Training loss')
# ‘b’는 파란색 실선을 의미합니다
plt.plot(epochs, val_loss, 'b', label='Validation loss')
plt.title('학습용, 평가용 데이터 - 손실')
plt.xlabel('Epochs')
plt.ylabel('Loss')
plt.legend()
plt.subplot(1,2,2)
acc = history_dict['accuracy']
val_acc = history_dict['val_accuracy']
plt.plot(epochs, acc, 'bo', label='Training acc')
plt.plot(epochs, val_acc, 'b', label='Validation acc')
plt.title('학습용, 평가용 데이터 - 정확도')
plt.xlabel('Epochs')
plt.ylabel('Accuracy')
plt.legend()
plt.show()
model = models.Sequential()
model.add(layers.Dense(16, activation='relu', input_shape=(10000,)))
model.add(layers.Dense(16, activation='relu'))
model.add(layers.Dense(1, activation='sigmoid'))
model.compile(optimizer=optimizers.RMSprop(learning_rate=0.001),
loss='binary_crossentropy',
metrics=['accuracy'])
model.fit(X_train, y_train, epochs=4, batch_size=128)
results = model.evaluate(X_test, y_test)
print("에폭수 : ", epochs, "배치사이즈 :", 128)
print("loss , accuracy", results)
Epoch 1/4 196/196 [==============================] - 2s 6ms/step - loss: 0.3465 - accuracy: 0.8613 Epoch 2/4 196/196 [==============================] - 1s 7ms/step - loss: 0.2074 - accuracy: 0.9223 Epoch 3/4 196/196 [==============================] - 1s 7ms/step - loss: 0.1677 - accuracy: 0.9380 Epoch 4/4 196/196 [==============================] - 1s 6ms/step - loss: 0.1435 - accuracy: 0.9485 782/782 [==============================] - 2s 2ms/step - loss: 0.3528 - accuracy: 0.8730 에폭수 : range(1, 21) 배치사이즈 : 128 loss , accuracy [0.3528025448322296, 0.8730400204658508]