csv 파일 정리 및 중간보고서 수정

ghdms
Commit e81467c6c315ebaeb5f45a50e8e7814257a172f0 e81467c6 1 parent 0772d864
Showing 6 changed files with 26 additions and 213 deletions
소스코드/kobis1.csv
소스코드/kobis2.csv
소스코드/movie.py
소스코드/movie_full.py
소스코드/save_kobis.py
진행보고서/중간보고서.docx
--- a/소스코드/kobis1.csv 0 → 100644
View file @e81467c
+++ b/소스코드/kobis1.csv 0 → 100644
View file @e81467c
--- a/소스코드/kobis2.csv deleted 100644 → 0
View file @0772d86
+++ b/소스코드/kobis2.csv deleted 100644 → 0
View file @0772d86
-명량,"17,615,658"
-극한직업,"16,265,855"
-신과함께-죄와 벌,"14,411,775"
-국제시장,"14,263,940"
-베테랑,"13,414,484"
-도둑들,"12,984,692"
-7번방의 선물,"12,811,714"
-암살,"12,706,829"
-"광해, 왕이 된 남자","12,324,062"
-신과함께-인과 연,"12,276,350"
-택시운전사,"12,189,654"
-부산행,"11,567,662"
-변호인,"11,375,123"
-해운대,"11,324,958"
-괴물,"10,917,400"
-왕의 남자,"10,513,976"
-기생충,"10,289,601"
-검사외전,"9,707,581"
-엑시트,"9,426,131"
-설국열차,"9,352,604"
-관상,"9,135,806"
-해적: 바다로 간 산적,"8,666,208"
-수상한 그녀,"8,660,623"
-백두산,"8,251,749"
-과속스캔들,"8,223,342"
-국가대표,"8,035,181"
-디워,"7,855,474"
-공조,"7,817,654"
-히말라야,"7,759,761"
-밀정,"7,500,457"
-최종병기 활,"7,470,633"
-써니,"7,363,139"
-1987,"7,232,387"
-베를린,"7,166,688"
-마스터,"7,150,586"
-터널,"7,120,780"
-내부자들,"7,073,516"
-인천상륙작전,"7,051,237"
-럭키,"6,975,631"
-은밀하게 위대하게,"6,959,083"
-범죄도시,"6,880,546"
-곡성,"6,879,989"
-화려한 휴가,"6,855,433"
-"좋은 놈, 나쁜 놈, 이상한 놈","6,686,075"
-늑대소년,"6,655,068"
-군함도,"6,592,151"
-웰컴 투 동막골,"6,436,900"
-사도,"6,247,651"
-아저씨,"6,178,665"
-미녀는 괴로워,"6,081,480"
-전우치,"6,065,474"
-연평해전,"6,045,049"
-타짜,"5,685,715"
-청년경찰,"5,653,444"
-숨바꼭질,"5,604,106"
-덕혜옹주,"5,599,995"
-더 테러 라이브,"5,584,295"
-감시자들,"5,509,019"
-검은 사제들,"5,443,232"
-안시성,"5,441,020"
-의형제,"5,416,923"
-더 킹,"5,318,007"
-완득이,"5,311,353"
-완벽한 타인,"5,294,154"
-타워,"5,181,244"
-투사부일체,"5,074,314"
-독전,"5,063,844"
-추격자,"5,046,096"
-공작,"4,975,517"
-바람과 함께 사라지다,"4,909,950"
-"님아, 그 강을 건너지 마오","4,802,416"
-봉오동 전투,"4,787,538"
-조선명탐정 : 각시투구꽃의 비밀,"4,786,259"
-군도: 민란의 시대,"4,775,811"
-남산의 부장들,"4,750,100"
-범죄와의 전쟁: 나쁜놈들 전성시대,"4,720,060"
-신세계,"4,684,571"
-도가니,"4,662,926"
-내 아내의 모든 것,"4,598,985"
-판도라,"4,585,116"
-나쁜 녀석들: 더 무비,"4,573,902"
-가문의 위기(가문의 영광2),"4,529,876"
-연가시,"4,516,063"
-강철비,"4,452,850"
-강철중: 공공의 적 1-1,"4,313,102"
-아가씨,"4,288,908"
-말아톤,"4,196,859"
-용의자,"4,131,338"
-건축학개론,"4,113,696"
-댄싱퀸,"4,058,225"
-7급 공무원,"4,039,891"
-우리 생애 최고의 순간,"4,018,872"
-꾼,"4,018,341"
-타짜-신의 손,"4,015,361"
-박수건달,"3,897,969"
-조선명탐정 : 사라진 놉의 딸,"3,872,015"
-역린,"3,849,700"
-남한산성,"3,849,129"
-암수살인,"3,789,553"
-국가부도의 날,"3,755,233"
-쌍화점,"3,749,034"
-신기전,"3,728,439"
-82년생 김지영,"3,679,019"
-귀향,"3,587,252"
-신의 한 수,"3,566,852"
-태풍,"3,471,150"
-부러진 화살,"3,460,212"
-끝까지 간다,"3,450,305"
-스파이,"3,436,092"
-그것만이 내 세상,"3,420,199"
-돈,"3,389,125"
-악인전,"3,364,712"
-이끼,"3,350,311"
-한반도,"3,331,593"
-포화속으로,"3,330,326"
-시동,"3,317,847"
-아이 캔 스피크,"3,283,470"
-마녀,"3,189,091"
-탐정: 리턴즈,"3,152,895"
-공공의 적 2,"3,134,424"
-퀵,"3,125,069"
-친절한 금자씨,"3,122,056"
-감기,"3,117,859"
-스물,"3,044,859"
-거북이 달린다,"3,024,666"
-헬로우 고스트,"3,019,960"
-하모니,"3,018,154"
-오싹한 연애,"3,009,406"
-방자전,"2,985,556"
-형,"2,982,514"
-마더,"2,981,953"
-그놈 목소리,"2,972,299"
-친구 2,"2,971,475"
-식객,"2,965,359"
-26년,"2,963,652"
-고지전,"2,945,151"
-워낭소리,"2,934,883"
-프리즌,"2,931,897"
-가장 보통의 연애,"2,924,563"
-말모이,"2,866,453"
-극비수사,"2,860,786"
-표적,"2,841,884"
-너의 결혼식,"2,820,969"
-내가 살인범이다,"2,729,827"
-부당거래,"2,723,773"
-소원,"2,711,094"
-너는 내 운명,"2,701,974"
-시라노; 연애조작단,"2,684,798"
-곤지암,"2,675,883"
-살인자의 기억법,"2,658,589"
-마파도,"2,636,857"
-후궁 : 제왕의 첩,"2,636,385"
-탐정 : 더 비기닝,"2,625,686"
-지금 만나러 갑니다,"2,602,273"
-가문의 부활 - 가문의 영광3,"2,598,859"
-위험한 상견례,"2,595,625"
-아수라,"2,594,695"
-보안관,"2,588,628"
-기술자들,"2,564,340"
-굿모닝 프레지던트,"2,554,960"
-태극기 휘날리며,"2,545,566"
-라스트 갓파더,"2,541,603"
-1번가의 기적,"2,535,450"
-증인,"2,534,793"
-목격자,"2,524,720"
-조작된 도시,"2,515,590"
-블랙머니,"2,480,145"
-반창꼬,"2,475,797"
-우리들의 행복한 시간,"2,472,160"
-조선명탐정: 흡혈괴마의 비밀,"2,444,136"
-화차,"2,436,884"
-이웃사람,"2,434,435"
-재심,"2,421,197"
-히트맨,"2,405,882"
-사바하,"2,398,581"
-화이: 괴물을 삼킨 아이,"2,394,487"
-의뢰인,"2,393,103"
-가문의 영광4 - 가문의 수난,"2,368,267"
-블라인드,"2,367,942"
-박열,"2,359,707"
-미인도,"2,342,579"
-음란서생,"2,304,368"
-내 생애 가장 아름다운 일주일,"2,290,709"
-하녀,"2,267,579"
-황해,"2,260,512"
-7광구,"2,242,510"
-타짜: 원 아이드 잭,"2,229,281"
-박쥐,"2,208,165"
-마당을 나온 암탉,"2,204,374"
-악의 연대기,"2,192,525"
-강남 1970,"2,192,276"
-신의 한 수: 귀수편,"2,159,081"
-마이 웨이,"2,142,670"
-나의 사랑 나의 신부,"2,141,155"
-바르게 살자,"2,135,606"
-내 사랑 내 곁에,"2,135,509"
-초능력자,"2,130,645"
-굿바이 싱글,"2,108,561"
-몽타주,"2,095,592"
-명당,"2,087,474"
--- a/소스코드/movie.py deleted 100644 → 0
View file @0772d86
+++ b/소스코드/movie.py deleted 100644 → 0
View file @0772d86
--- a/소스코드/movie_full.py
View file @e81467c
+++ b/소스코드/movie_full.py
View file @e81467c
@@ -5,9 +5,14 @@ conn = pymongo.MongoClient("localhost")
 db = conn.test
 MOVIE = db.movie
 cur = MOVIE.find()
-audience = {}
+audiences = {}
+salesPerScreens = []
 for c in cur:
-    audience[c["name"]] = c["audience"]
+    audiences[c["name"]] = c["audiences"]
+    salesPerScreens.append({"name": c["name"], "value": c["sales"] / c["screens"]})
+
+salesPerScreensTmp = salesPerScreens[:]
+salesPerScreensSorted = sorted(salesPerScreensTmp, key=lambda n: salesPerScreens[salesPerScreens.index(n)]["value"], reverse=True)[:100]
 data = {
     "명량": "최민식,류승룡,조진웅,김명곤,진구,이정현,권율,노민우,이승준,박보검,김원해,이해영,장준녕,김길동,김강일,강태영,박노식,최덕문,조복래,문영동,이주실,주석태,고경표,김태훈,한이진,최창균,신창수,김현태,정재우,김구택,이재구,조하석",
@@ -36,6 +41,7 @@ data = {
     "백두산": "이병헌,하정우,마동석,전혜진,배수지,이상원,옥자연,한수현,강신철,이경영,조한철,최광일,김시아,임형국,박주환,남성준,서현우,김민식,박지홍,김준원,남문철,박성근",
     "과속스캔들": "차태현,박보영,왕석현,임지규,황우슬혜,임승대,정원중,김기방,박영서",
     "국가대표": "하정우,성동일,김지석,김동욱,최재환,이재응,이은성,이한위,이혜숙,김지영,현쥬니,황하나",
+    #"디워": "", #외국 배우
     "공조": "현빈,유해진,김주혁,장영남,이해영,윤아,이동휘,공정환,엄효섭,신현빈,박민하,전국환,이이경,박진우,박형수",
     "히말라야": "황정민,정우,조성하,김인권,라미란,김원해,이해영,전배수,정규수,성병숙",
     "밀정": "송강호,공유,한지민,엄태구,신성록,허성태,서영주,정유안,이수광,고준,최유화,김동영,곽자형,조영규,이설구,권수현,이환,유상재,한수연,남문철,김수웅,최장원,허형규,백인권,정도원,신성일,김의건",
@@ -79,7 +85,7 @@ data = {
     "추격자": "김윤석,하정우,서영희,박효주,김유정,최정우,정인기,구본웅,민경진",
     "공작": "황정민,이성민,조진웅,주지훈,김홍파,정소리,기주봉,김응수,채용,박진영,남문철,최병모,김인우,정기섭,이승준,양현민,허승,엄지만,백승익,권범택,최정인,김소진,박혜영,김지나,곽자형,조주희",
     "바람과 함께 사라지다": "차태현,오지호,민효린,성동일,신정근,고창석,송종호,이채영,김길동,김향기,천보근,김구택,독고준",
-#    "님아, 그 강을 건너지 마오": "", #영화배우가 아님
+    #"님아, 그 강을 건너지 마오": "", #영화배우가 아님
     "봉오동 전투": "유해진,류준열,조우진,박지환,최유화,성유빈,이재인,홍상표,원풍연,양현민",
     "조선명탐정 : 각시투구꽃의 비밀": "김명민,한지민,오달수,최무성,우현,이재용,예수정,정인기",
     "군도: 민란의 시대": "하정우,강동원,조진웅,마동석,정만식,김성균,주진모,이경영,이성민,윤지혜,김재영,송영창,김병옥,임현성,강현중",
@@ -145,7 +151,7 @@ data = {
     # "식객": "",
     # "26년": "",
     # "고지전": "",
-    # "워낭소리": "",
+    "워낭소리": "",
     # "프리즌": "",
     # "가장 보통의 연애": "",
     # "말모이": "",
@@ -155,11 +161,11 @@ data = {
     # "내가 살인범이다": "",
     # "부당거래": "",
     # "소원": "",
-    # "너는 내 운명": "",
+    "너는 내 운명": "",
     # "시라노; 연애조작단": "",
     # "곤지암": "",
     # "살인자의 기억법": "",
-    # "마파도": "",
+    "마파도": "이정진,이문식,여운계,김수미,김을동,김형자,길해연,오달수,서영희",
     # "후궁 : 제왕의 첩": "",
     # "탐정 : 더 비기닝": "",
     # "지금 만나러 갑니다": "",
@@ -169,7 +175,7 @@ data = {
     # "보안관": "",
     # "기술자들": "",
     # "굿모닝 프레지던트": "",
-    # "태극기 휘날리며": "",
+    "태극기 휘날리며": "장동건,원빈,이은주,공형진,장민호,이영란",
     # "라스트 갓파더": "",
     # "1번가의 기적": "",
     # "증인": "",
@@ -232,6 +238,10 @@ def named_union(graph1, graph2): #두 그래프 합성
     Z.vs["label"] = Z.vs["name"][:]
     return Z
+for i in range(0, 100):
+    if(salesPerScreensSorted[i]["name"] not in data):
+        print(i, salesPerScreensSorted[i])
+
 frequency = {}
 weight = {}
 keys = []
@@ -263,12 +273,12 @@ for movie in data:
             join = join[0]+","+join[1]
             if(join not in weight):
                 weight[join] = {
-                "audience": audience[movie],
+                "audiences": audiences[movie],
                 "count": 1
                 }
                 keys.append(join)
             else:
-                weight[join]["audience"] += audience[movie]
+                weight[join]["audiences"] += audiences[movie]
                 weight[join]["count"] += 1
 topten = sorted(keys, key=lambda n: weight[n]["count"], reverse=True)[:10]
--- a/소스코드/save_kobis.py
View file @e81467c
+++ b/소스코드/save_kobis.py
View file @e81467c
@@ -6,9 +6,12 @@ db = conn.test
 MOVIE = db.movie
 MOVIE.delete_many({}) #reset
-csvfile = open('kobis2.csv', mode='r')
+csvfile = open('C:\\Users\\ghdms\\2014104137\\소스코드\\kobis1.csv', mode='r')
 reader = list(csv.reader(csvfile))
 for row in reader:
-    num = int(row[1].replace(",", ""))
+	name = row[1]
-    print(num)
+	for i in range(2, 6):
-    MOVIE.insert_one({"name": row[0], "audience": num})
+		row[i] = row[i].replace(",", "")
+	sales, audiences, screens, plays = int(row[2]), int(row[3]), int(row[4]), int(row[5])
+	print(name, sales, audiences, screens, plays)
+	MOVIE.insert_one({"name": name, "sales": sales, "audiences": audiences, "screens": screens, "plays": plays})
\ No newline at end of file
--- a/진행보고서/중간보고서.docx
View file @e81467c
+++ b/진행보고서/중간보고서.docx
View file @e81467c