Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
모빌리티데이터팀 신입 데이터 분석가의 1년 회고
Search
SOCAR
October 19, 2019
Programming
0
870
모빌리티데이터팀 신입 데이터 분석가의 1년 회고
데이터야놀자 2019에서 발표한 자료입니다
SOCAR
October 19, 2019
Tweet
Share
More Decks by SOCAR
See All by SOCAR
코드 리뷰 적응기
socar
0
880
O'reilly Strata Data Conference New York 2019 후기
socar
2
1.4k
TF에서 팀 빌딩까지 9개월의 기록 : 성장하는 조직을 만드는 여정
socar
0
290
쏘카, 개발 - 쏘카 R&D 이야기
socar
0
440
MOBILITY x DATA : 모빌리티 산업의 도전 과제
socar
0
590
Other Decks in Programming
See All in Programming
Pinia Colada が実現するスマートな非同期処理
naokihaba
4
220
ペアーズにおけるAmazon Bedrockを⽤いた障害対応⽀援 ⽣成AIツールの導⼊事例 @ 20241115配信AWSウェビナー登壇
fukubaka0825
6
1.9k
色々なIaCツールを実際に触って比較してみる
iriikeita
0
330
3rd party scriptでもReactを使いたい! Preact + Reactのハイブリッド開発
righttouch
PRO
1
600
ふかぼれ!CSSセレクターモジュール / Fukabore! CSS Selectors Module
petamoriken
0
150
ActiveSupport::Notifications supporting instrumentation of Rails apps with OpenTelemetry
ymtdzzz
1
230
Remix on Hono on Cloudflare Workers
yusukebe
1
290
Creating a Free Video Ad Network on the Edge
mizoguchicoji
0
120
PHP でアセンブリ言語のように書く技術
memory1994
PRO
1
170
Compose 1.7のTextFieldはPOBox Plusで日本語変換できない
tomoya0x00
0
190
3 Effective Rules for Using Signals in Angular
manfredsteyer
PRO
1
100
AWS IaCの注目アップデート 2024年10月版
konokenj
3
3.3k
Featured
See All Featured
How to train your dragon (web standard)
notwaldorf
88
5.7k
Performance Is Good for Brains [We Love Speed 2024]
tammyeverts
6
410
Designing for Performance
lara
604
68k
Creating an realtime collaboration tool: Agile Flush - .NET Oxford
marcduiker
25
1.8k
CSS Pre-Processors: Stylus, Less & Sass
bermonpainter
356
29k
Imperfection Machines: The Place of Print at Facebook
scottboms
265
13k
How STYLIGHT went responsive
nonsquared
95
5.2k
Why Our Code Smells
bkeepers
PRO
334
57k
Site-Speed That Sticks
csswizardry
0
23
Practical Tips for Bootstrapping Information Extraction Pipelines
honnibal
PRO
10
720
Git: the NoSQL Database
bkeepers
PRO
427
64k
Fireside Chat
paigeccino
34
3k
Transcript
ݽ࠽ܻ౭ ؘఠ नੑ ࠙ࢳо 1֙ ഥҊ
ࣗѐ ӂਮജ • ҃ + ஹೊఠҕ ҕ ݏ ঋই ز
• झఋসҗ ؘఠী ҙब ݆ • 2019.01 ॑ ؘఠ ࠙ࢳо ੑࢎ, ఋ ؘఠ ѐੋੋ ҙब • بद ࢤ, بद ӝמ ١ بदী ҙब ݆ • নೠ بदܳ ҃ೞח Ѫਸ જইೣ • ࢲ݅ఀ Ѣೠ بदח ইࠁӝ ൨ٜ ఋؘఠ • ର ബਯച • ࣻਃ ஏ, ର ߓ ঌҊ્ܻ • ࢲ زച
ز җѢ৬ അ
ܻо زೞח ߑध ӝࣿ ߊҗ ೣԋ ߸ച
ؘఠ۽ ࠁח ࢲ ز ؘఠ۽ ಹח ݽ࠽ܻ౭ ޙઁ
ഥҊ ETA ର ബਯച
1. ؘఠ۽ ࠁח ࢲ ز
ৈ۞ٜ࠙য٣ࢲয়࣑աਃ
ࢲزীҙೠݻоࢎपٜ
1 ъթҳ 2 ҳ 3 ࢲୡҳ 4 ١ನҳ 5 ઙ۽ҳ
6 ਊҳ 7 ݃ನҳ 8 ࣠ҳ 9 ࢿزҳ 10 ࢲޙҳ 11 ҙঈҳ 12 ъࢲҳ 0.0147 13 زҳ 0.0135 14 ࢿ࠘ҳ 0.0123 15 ҳ۽ҳ 0.0106 ইஜ 8:00 ب ࣽਤ (%) ইஜীࢎۈٜೱೞחҔ
ইஜ 8:00 ب ࣽਤ (%) ইஜীࢎۈٜೱೞחҔ ب ࢚ਤ 5ѐ ҳח
ز ݻ %ܳ ରೡөਃ? ࢲ ҳח ୨ 25ѐ ੑפ. 1 ъթҳ 2 ҳ 3 ࢲୡҳ 4 ١ನҳ 5 ઙ۽ҳ 6 ਊҳ 7 ݃ನҳ 8 ࣠ҳ 9 ࢿزҳ 10 ࢲޙҳ 11 ҙঈҳ 12 ъࢲҳ 0.0147 13 زҳ 0.0135 14 ࢿ࠘ҳ 0.0123 15 ҳ۽ҳ 0.0106
ইஜ 8:00 ب ࣽਤ (%) 24% 76% ইஜীࢎۈٜೱೞחҔ 1 ъթҳ
0.3041 2 ҳ 0.1353 3 ࢲୡҳ 0.1324 4 ١ನҳ 0.0874 5 ઙ۽ҳ 0.0784 6 ਊҳ 0.0657 7 ݃ನҳ 0.0369 8 ࣠ҳ 0.0318 9 ࢿزҳ 0.0282 10 ࢲޙҳ 0.0275 11 ҙঈҳ 0.0213 12 ъࢲҳ 0.0147 13 زҳ 0.0135 14 ࢿ࠘ҳ 0.0123 15 ҳ۽ҳ 0.0106
ࢲबসޖҳ $#%$FOUSBM#VTJOFTT%JTUSJDU ١ನ ҳ ઙ۽ ъթ
↟بद࢚সӝמػٜ ↟݆زबসޖҳܳबਵ۽ੌযդ ١ನ ъթ ҳ ઙ۽ ࢲबসޖҳ $#%$FOUSBM#VTJOFTT%JTUSJDU
١ನҳ Ӕ ࢲ թࢲଃী ݽৈ
ҳ,ઙ۽ҳ Ӕ ੌఠܳ ل۞ऱҊ ࢎۈٜ Ҋ
ъթҳ Ӕ ъթ সޖҳ৬ Ѣо ࢴৈח ݽण
बঠ ۽оࠇद
ࢎۈٜয٣ࢲೞܖܳ݃ޖܻೡө ঠр 23:00 ഐ ࣽਤ (%) 1 ъթҳ 2 ҳ
3 ࢲୡҳ 4 ઙ۽ҳ 5 ਊҳ 6 ݃ನҳ 7 ١ನҳ 8 ࢿزҳ 9 ࣠ҳ 10 ࢲޙҳ 11 ҙঈҳ 0.0109 12 زҳ 0.0101 13 ҟҳ 0.0079 14 ҳ۽ҳ 0.0065 15 ъࢲҳ 0.0064 ഐ ࢚ਤ 5ѐ ҳח ز ݻ %ܳ ରೡөਃ?
ࢎۈٜয٣ࢲೞܖܳ݃ޖܻೡө ঠр 23:00 ഐ ࣽਤ (%) 1 ъթҳ 0.4241 2
ҳ 0.1248 3 ࢲୡҳ 0.0982 4 ઙ۽ҳ 0.0791 5 ਊҳ 0.0783 6 ݃ನҳ 0.052 7 ١ನҳ 0.0435 8 ࢿزҳ 0.0216 9 ࣠ҳ 0.0159 10 ࢲޙҳ 0.0146 11 ҙঈҳ 0.0109 12 زҳ 0.0101 13 ҟҳ 0.0079 14 ҳ۽ҳ 0.0065 15 ъࢲҳ 0.0064 20% 80%
None
ࢎۈٜয٣ࢲೞܖܳ݃ޖܻೡө ঠр 23:00 ഐ ࣽਤ (%) 1 ъթҳ 0.1897 2
ࢲୡҳ 0.1086 3 ਊҳ 0.0817 4 ҙঈҳ 0.0741 5 ࣠ҳ 0.0694 6 ࢿزҳ 0.0694 7 ݃ನҳ 0.0667 8 زҳ 0.0607 9 ١ನҳ 0.0531 10 ҟҳ 0.0431 11 ҳ 0.0406 12 ࢿ࠘ҳ 0.0398 13 ࢲޙҳ 0.0379 14 ಣҳ 0.0339 15 ઙ۽ҳ 0.0312 ঠр 23:00 ب ࣽਤ (%) 1 ъթҳ 0.4241 2 ҳ 0.1248 3 ࢲୡҳ 0.0982 4 ઙ۽ҳ 0.0791 5 ਊҳ 0.0783 6 ݃ನҳ 0.052 7 ١ನҳ 0.0435 8 ࢿزҳ 0.0216 9 ࣠ҳ 0.0159 10 ࢲޙҳ 0.0146 11 ҙঈҳ 0.0109 12 زҳ 0.0101 13 ҟҳ 0.0079 14 ҳ۽ҳ 0.0065 15 ъࢲҳ 0.0064
बঠदр1. ߊ ب
ইஜदр". ߊ ب
ݽ࠽ܻ౭ؘఠ࠙ࢳоח۽যڃؘఠܳࠁաਃ Origin - Destination data • ౠ दр زউ যו
ࢎۈٜ য٣۽ زೞӡ ਗೞחо? • ࠂೠ بद ࣘ ز ಁఢী ೠ Ѫ
ݽ࠽ܻ౭ؘఠ࠙ࢳоח۽যڃؘఠܳࠁաਃ Origin - Destination data • ؘఠࣁ द id created_at_kr
status origin_lng origin_lat origin_gu destination_lng destination_lat dest_si dest_gu 1 2019-01-30T00:25:20 CANCELED 127.0414496 37.51002284 ъթҳ 127.1909267 37.56209539 ҃ӝ ೞթद 2 2019-011-27T15:37:40 ACCEPTED 127.0563954 37.54280202 ࢿزҳ 126.8721171 37.45949251 ҃ӝ ҟݺद 3 2019-09-16T17:52:19 ARRIVED_AT 127.0613762 37.51048997 ъթҳ 127.1375613 37.59445381 ҃ӝ ҳܻद 4 2019-04-22T19:43:07 PICKED_UP 127.1105661 37.51265727 ࣠ҳ 126.9714429 37.40572274 ҃ӝ উনद 5 2019-07-07T21:33:56 RIDING 127.0127873 37.49320677 ࢲୡҳ 127.0709374 37.27701722 ҃ӝ ਊੋद 6 2019-10-30T21:35:33 DROPPED_OFF 127.0018494 37.58132881 ࢿزҳ 127.1243009 37.32402575 ҃ӝ द 7 2019-09-27T21:41:36 DISPATCHING 127.0730746 37.5113235 ࣠ҳ 126.983202 37.39739958 ҃ӝ ৴द 8 2019-12-24T22:53:23 CANCELED 126.9696439 37.5635487 ҳ 126.9936623 37.43394209 ҃ӝ җୌद ഐ / ਤ҃ب ب / ਤ҃ب दр
ݽ࠽ܻ౭ؘఠ࠙ࢳоח۽যڃؘఠܳࠁաਃ Origin - Destination data • ؘఠࣁ द id created_at_kr
status origin_lng origin_lat origin_gu destination_lng destination_lat dest_si dest_gu 1 2019-01-30T00:25:20 CANCELED 127.0414496 37.51002284 ъթҳ 127.1909267 37.56209539 ҃ӝ ೞթद 2 2019-011-27T15:37:40 ACCEPTED 127.0563954 37.54280202 ࢿزҳ 126.8721171 37.45949251 ҃ӝ ҟݺद 3 2019-09-16T17:52:19 ARRIVED_AT 127.0613762 37.51048997 ъթҳ 127.1375613 37.59445381 ҃ӝ ҳܻद 4 2019-04-22T19:43:07 PICKED_UP 127.1105661 37.51265727 ࣠ҳ 126.9714429 37.40572274 ҃ӝ উনद 5 2019-07-07T21:33:56 RIDING 127.0127873 37.49320677 ࢲୡҳ 127.0709374 37.27701722 ҃ӝ ਊੋद 6 2019-10-30T21:35:33 DROPPED_OFF 127.0018494 37.58132881 ࢿزҳ 127.1243009 37.32402575 ҃ӝ द 7 2019-09-27T21:41:36 DISPATCHING 127.0730746 37.5113235 ࣠ҳ 126.983202 37.39739958 ҃ӝ ৴द 8 2019-12-24T22:53:23 CANCELED 126.9696439 37.5635487 ҳ 126.9936623 37.43394209 ҃ӝ җୌद ഐ ஂࣗ ٘ۄߡ ഐ ࣻۅ ٘ۄߡ ب थё थ ೯ थё ೞର ӝ ࢚క 다양한 상태 값을 가짐
2. ؘఠ۽ ಹח ݽ࠽ܻ౭ ޙઁ
https://dribbble.com/shots/4153309-The-Wait ޖоӝܻחܻݽण jj
ରഐ ରب ݾߊ ݾب Icons made by Freepik from www.flaticon.com
is licensed by CC 3.0 BY ఋ҃ৈ
ରഐ ରب ݾߊ ݾب Icons made by Freepik from www.flaticon.com
is licensed by CC 3.0 BY ఋ҃ৈ 데이터로 사용자 경험을 개선한 이야기
ETA ۆ? • Estimated time of arrival ( ࢚ ب
दр ) • ରਸ ഐೞҊ Ҋёীѱ بೞӝө ݽ࠽ܻ౭ী.-ਊೞӝ
ݽ࠽ܻ౭ী.-ਊೞӝ 주요 호출 취소 사유
&5"ޙઁܳ೧Ѿೞחߑߨٜ ରਸט۰ࢲ Әߑبೞѱ݅ٚ
ରਸט۰ࢲ Әߑبೞѱ݅ٚ تҗदрٚ അपੋޙઁ &5"ޙઁܳ೧Ѿೞחߑߨٜ
&5"ޙઁܳ೧Ѿೞחߑߨٜ ࠙ױਤ۽߈ৢܿਸೠ ରਸט۰ࢲ Әߑبೞѱ݅ٚ تҗदрٚ അपੋޙઁ
&5"ޙઁܳ೧Ѿೞחߑߨٜ ࠙ੋؘ࠙ਵ۽ ցޖࠗഛೞ ࠙ױਤ۽߈ৢܿਸೠ ରਸט۰ࢲ Әߑبೞѱ݅ٚ تҗदрٚ അपੋޙઁ
&5"ޙઁܳ೧Ѿೞחߑߨٜ ࠙ੋؘ࠙ਵ۽ ցޖࠗഛೞ ࠙ױਤ۽߈ৢܿਸೠ ରਸט۰ࢲ Әߑبೞѱ݅ٚ تҗदрٚ അपੋޙઁ ؊ഛೠчਸઁҕ೧ঠೠ
&5"ޙઁܳ೧Ѿೞחߑߨٜ ݠन۞ ݽ؛ਸ ਊ೧ࠁ ࠙ੋؘ࠙ਵ۽ ցޖࠗഛೞ ࠙ױਤ۽߈ৢܿਸೠ ରਸט۰ࢲ Әߑبೞѱ݅ٚ
تҗदрٚ അपੋޙઁ ؊ഛೠчਸઁҕ೧ঠೠ
۽ંݾ ↟Ҋёীѱࠁഛೠ࢚بदрਸࠁৈ ↟ఎਸ݄Ҋ ࢲ࠺झਸ֫חੌ ӝઓ&5" ֎ߡ"1* ࠁ&5" .-ݽ؛ ↟ ٘ۄߡҗѢ೯ಁఢ
&5"ҙ۲ؘఠఐ࢝ दрীٮܲ&5"۪٘ ਘ߹ दр߹ 6ਘ ର ૐର റ хࣗ ୶ࣁ
Ӕ दр ృӔ दр
&5"ҙ۲ؘఠఐ࢝ ীٮܲಣӐ&5" Aҳ Bҳ
٘ۄߡী ٮܲ ಞରо &5"ҙ۲ؘఠఐ࢝ ࢚ ب दрࠁ ןח ҃ೱࢿ
= SUM( ۄ٘ Ѥٜ पઁ ࣗਃ दр) / SUM( ۄ٘ Ѥٜ ࢚ ࣗਃ दр)
٘ۄߡী ٮܲ ಞରо &5"ҙ۲ؘఠఐ࢝
٘ۄߡী ٮܲ ಞରо &5"ҙ۲ؘఠఐ࢝ ডр ןח ࢿೱ 1.0 ~
1.5
٘ۄߡী ٮܲ ಞରо &5"ҙ۲ؘఠఐ࢝ ן ঋח ࢿೱ
ࢎਊೠೖٜ ٘ۄߡןח҃ೱࢿ Ҋёডஂࣗా҅ ӝઓ&5" दр ਃੌ ର_ഐࢎѢܻ ਤب ҃ب ߸ࣻ
ਃب
ࢎਊೠೖٜ ٘ۄߡןח҃ೱࢿ Ҋёۄ٘ஂࣗా҅ ޖоࡐݡѪэו՝
ࢎਊೠೖٜ ٘ۄߡןח҃ೱࢿ Ҋёۄ٘ஂࣗా҅ ղо݅ٚݽ؛࠙ੌନয়חѪҗ࠙ੌନןѱѪਸҳ࠙ೞޅೣ അपੌନয়חѪҗןѱয়חѪೱܰ
ݽ࠽ܻ౭ী.-ਊೞӝ ಣо • ੌ߈ਵ۽ ॳח MSE ࣚप ೣࣻ۽
णೠݶ • Ҋёীѱ ןח Ѫҗ ࡈܻ য়ח Ѫਸ ڙэ ੋध
ݽ࠽ܻ౭ী.-ਊೞӝ ಣо • Ҋё ҃ী ؊ աࢂ ೱਸ
ח Ѫ ןѱ য়ח Ѫ • ٮۄࢲ ןѱ য়ח Ѫী ಕօ౭ܳ ؊ ب۾ • ࣚप ೣࣻܳ • ࠺ட ਤ (Asymmetrical risk) • ੌ߈ਵ۽ ॳח MSE ࣚप ೣࣻ۽ णೠݶ • Ҋёীѱ ןח Ѫҗ ࡈܻ য়ח Ѫਸ ڙэ ੋध
ݽ࠽ܻ౭ী.-ਊೞӝ ಣо
ݽ࠽ܻ౭ী.-ਊೞӝ ಣо weight = 0.4 residual = y_true
- y_predict grad = np.where(residual<0, -2.0*residual, -2.0*weight*residual) য়ରо 0ࠁ ҃৬ ҃ী ಁօ౭ܳ ܰѱ ח ೣࣻܳ Params = { ‘objective’ : custom_asymmetric_objective } ݾೣࣻܳ ೠ ೣࣻ۽
ݽ࠽ܻ౭ী.-ਊೞӝ Ѿҗ
ݽ࠽ܻ౭ী.-ਊೞӝ Ѿҗ
ݽ࠽ܻ౭ী.-ਊೞӝ Ѿҗ
ݽ؛݂ ױ҅ীࢲ ו՛ Ѫ • അपী ਊೡ ٸח নೠ ٜਸ
ೣԋ Ҋ೧ঠೠ • ఋ ࢎ۹ • ഛب : MAE • ࠺פझ : п ࠺ਯਸ ծ୶ח Ѫ / Ҋё ҃ ೱ࢚ ഛب ࠺פझ ҙ ݽ࠽ܻ౭ী.-ਊೞӝ
ߓನ ળ࠺ ױ҅ ݽ࠽ܻ౭ী.-ਊೞӝ
ߓನ ળ࠺ ױ҅ • ݽ؛݂ ৮߷ೞѱ լ • ઁ ࣁ࢚ਵ۽
ղࠁղ ݽ࠽ܻ౭ী.-ਊೞӝ ࠁ ݽ؛ ӝઓ ETA ࠁ ETA
ߓನ റ ݽ࠽ܻ౭ী.-ਊೞӝ
ঌ ࣻ হח ਗੋ ݽ࠽ܻ౭ী.-ਊೞӝ • যڃ ࠙ನ ೖٜ ٜয৳חী
ೠ ۽Ӧਸ ೞ ঋও ࠁ ݽ؛ ӝઓ ETA ࠁ ETA
ঌ ࣻ হח ਗੋ ݽ࠽ܻ౭ী.-ਊೞӝ ࠗपೠ ۽Ӧ • पઁ ч
• ஏ ч • যڃ ࠙ನ ೖٜ ٜয৳חী ೠ ۽Ӧਸ ೞ ঋও • ٮۄࢲ ݽ؛ ৵ Ӓۧѱ ஏ೮ח ঌ ࣻо হ ࠁ ݽ؛ ӝઓ ETA ࠁ ETA
ੌױ ܀ߔ ݽ࠽ܻ౭ী.-ਊೞӝ • ٘ۄ۠ - ࢲ࠺झীח ਊೞ ঋҊ पઁ
ജ҃ীࢲ పझ݅ ೯ ࠁ ݽ؛ ӝઓ ETA ࠁ ETA ӝઓ ETA
ੌױ ܀ߔ ݽ࠽ܻ౭ী.-ਊೞӝ • ٘ۄ۠ - ࢲ࠺झীח ਊೞ ঋҊ पઁ
ജ҃ীࢲ పझ݅ ೯ • ࠻௪ܻ৬ క࠶۽۽ ݽפఠ݂ بҳܳ ݃۲ೞҊ ۽Ӓܳ ऺইࠆ ࠁ ݽ؛ ӝઓ ETA ࠁ ETA ӝઓ ETA
য়ܨо լ؍ ਗੋ ݽ࠽ܻ౭ী.-ਊೞӝ • ਃೠ ೖ ೞաо ࢲ۽
ܲ ࠙ನܳ оҊ (ࢲ۽ ܲ ױਤ ޙઁ) Train ױ҅
য়ܨо լ؍ ਗੋ ݽ࠽ܻ౭ী.-ਊೞӝ • ਃೠ ೖ ೞաо ࢲ۽
ܲ ࠙ನܳ оҊ (ࢲ۽ ܲ ױਤ ޙઁ) Train ױ҅ Test ױ҅
য়ܨо լ؍ ਗੋ ݽ࠽ܻ౭ী.-ਊೞӝ • ਃೠ ೖ ೞաо ࢲ۽
ܲ ࠙ನܳ о • ߓನ ী ݽפఠ݂ਸ ࠺೮ݶ Әߑ ೧Ѿ೮ਸ ޙઁ • ѐߊ৬ ഈস җীࢲ ࢎ ࣗా ޙઁ Train ױ҅ Test ױ҅
ݽ࠽ܻ౭ী.-ਊೞӝ য ؘఠࣇ ࠙ࢳ ߂ ݽ؛݂ য ಣо
(MAE,RMSE) ঌ؍ Ѫ
য ؘఠࣇ ࠙ࢳ ߂ ݽ؛݂ য ಣо (MAE,RMSE)
࠙ࢳ ߂ ݽ؛݂ ࢜۽ ೞח ಣо ࢜۽ ݅٘ח ؘఠࣇ ߓನ ߂ ݽפఠ݂ ঌ؍ Ѫ ೧ঠ೮؍ Ѫ ݽ࠽ܻ౭ী.-ਊೞӝ
2. ؘఠ۽ ಹח فߣ૩ ݽ࠽ܻ౭ ޙઁ
٘ۄߡоغযࠇद ࢚ടীࢲ ѐੋ ղܾ ࣻ ח ୭ ࢶఖ? ࢚
ࣻਃ
٘ۄߡоغযࠇद ѐੋ ౸ױೡ ࣻ ח ୭ ࢶఖ ࢚ ࣻਃ
Ӓۢ٘ۄߡоৈ۞ݺۄݶ ࢚ ࣻਃ
࢚ ࣻਃ بद ҙীࢲח ୭ ইצ Ѿҗ Ӓۢ٘ۄߡоৈ۞ݺۄݶ
ҙ 'MFFU ীࢲ୭ࢶఖ ࢚ ࣻਃ
ҙ 'MFFU ীࢲ୭ࢶఖ ࢚ ࣻਃ
ҙ 'MFFU ীࢲ୭ࢶఖ ࢚ ࣻਃ ରബਯച
ҙ 'MFFU ীࢲ୭ࢶఖ ࢚ ࣻਃ ରബਯച ݠन۞दр߹߹ࣻਃPSҕәਸஏ ୭ചݽ؛ஏೠчܳ߄ఔਵ۽୭ച +
• Linear Programming • Integer Programming • Routing • Packing
• Network Flows • Assignment • Scheduling Operation research
• Linear Programming • Integer Programming • Routing • Packing
• Network Flows • Assignment • Scheduling Operation research
ݽ࠽ܻ౭৬୭ച
֎ਕ • ֢٘৬ ݂۽ ܖয ֎ਕ
֢٘ /PEF ೞӝ -20 -20 -20 -10 -50 -40 90
30 • ֢٘ח ౠೠ ۄҊ ೡ ࣻ • ҕә ֢٘ : թח ର ߊࢤೞח • ࣻਃ ֢٘ : ࣻਃо ߊࢤೞח
݂ -JOL ೞӝ • ݂ח ֢٘৬ ֢٘ܳ োѾೠ • ݂ܳ
ా೧ زਸ ೡ ҃ীח ࠺ਊ ߊࢤ • അपীࢲח Ѣܻ ࣗਃ दр 30 5 10 12 12 20 5 10
-20 -20 -20 -10 -50 -40 90 30 • ܻо
ೠ ֎ਕ 30 5 10 12 12 20 5 10 ֎ਕ
ݽ࠽ܻ౭৬୭ച ݂ ࠺ਊਸ п ֢٘ ҕәҗ ࣻਃܳ ࣛߡܳ
ࢶೞҊ, ֎ਕܳ ҳࢿ https://developers.google.com/optimization/flow/mincostflow
-20 -10 -50 -40 90 30 ݽ࠽ܻ౭৬୭ച • ࠺ਊਸ ୭ࣗചೞݶࢲ
ҕә ֢٘ীࢲ ࣻਃ ֢٘۽ زदఃח ߑध • Min cost flow problem
ݽ࠽ܻ౭৬୭ച 0 4 5 3 1 2 20 10 50
40 50 • ࣛߡ(Solver)о ળ
ݽ࠽ܻ౭৬୭ച 0 4 5 3 1 2 20 10 50
40 50 • ࣛߡ(Solver)о ળ • അपীࢲ Ҋ۰೧ঠೡ ࢎ೦ٜ • ز ী ൧ ࣻب • ҕә োࣘਵ۽ ߊࢤೠ
3. ഥҊ
.-ݽ؛ਸ݅٘חѪҗपઁജ҃ীࢲ࠺झೞחѪ ݽ؛݂ Ӓ റܳ ࢤпೞӝ • ߓನ റীب উਵ۽ ҙܻೡ
ࣻ ח ജ҃ਸ ݃۲೧فח Ѫ ਃ (ݽפఠ݂) ࢚ടী ݏח ݾೣࣻ ӝ • ਃೞݶ ࣚप ೣࣻܳ ೧ঠೡ ࣻب • Objective is subjective !
بݫੋীҙೠঠӝ ־ҳաীѱ ਸ п بݫੋ • ೧ بݫੋীࢲח ݠन۞, ٩۞
ࠗо ইק ࣻ Ҋ • ܲ ӝߨٜҗ ೣԋ ॳৈঠ ࡄਸ ߊೡ ٸо ח Ѫ э
ۨਃࢿ ഒࢲ ݽٚ Ѫਸ ೡ ࣻח হӝী • ؘఠূפয +
ؘఠ࠙ࢳо + ѐߊ + • নೠ ҵ ࢎۈٜҗ ഈসਵ۽ ޙઁܳ ೧Ѿ೧աоҊ ߸ਵ۽ࠗఠ ݆ ߓӝ • જ ࢎࣻ, ܲ ҵ زܐ
ߊܳب৬दҊ ೦࢚хਸחLZMF https://zzsza.github.io/
ਊ ؘఠӒܛ • ఋؘఠ • ؘఠࢎझ • оѺۚ • ؘఠূפয݂
• URL : bit.ly/ؘఠঠ֥॑ਊ ؘఠ • ؘఠূפয • ؘఠࢎ౭झ • URL : https://tadacareer.vcnc.co.kr/
хࢎפ