클러스터 분석

UDIS 6th Study

Transcript

1. ௿۞झఠ ࠙ࢳ
   Cluster Analysis Sean Lee @ UDIS 6th Study - ؘ੉ఠ ӝ߈ ࢲ࠺झ ٣੗ੋ https://accounts-flickr.yahoo.com/photos/karakola/3234501973/?rb=1
2. ؘ੉ఠ • How? • ా҅೟੸ ੽Ӕ • ؘ੉ఠ Mining
3. ؘ੉ఠ Mining Advanced Mining Opinion Mining Social Network Mining Cluster Mining https://www.flickr.com/photos/stavos52093/9645884201/
4. Advanced Mining Decision Tree, Machine Learning, … ??? Opinion Mining ࢎۈٜ੄ ੄Ѽ੉ա కبܳ ࠙ࢳ / ౟ਤఠ ৈۿ ࠙ࢳ Social Network Mining ҙ҅ܳ ాೠ ࠙ࢳ / ӂ۱૑ഋب, Ҵ੿ਗ؆Ӗ҅੿Ѩ࢝ Cluster Mining ਬࢎೠ ؘ੉ఠܳ ޘযࢲ ࠙ࢳ / ಕ੉झ࠘ ஘ҳ୶ୌ
5. ௿۞झఠ Mining(Analysis)
7. ௿۞झఠ ҵ૘ ޘ਺ ׮ߊ
8. https://www.flickr.com/photos/skiwalker79/3306092836/
9. ௿۞झఠ ࠙ࢳ • ࠺तೠ(similar) ؘ੉ఠܳ 
 ޘযࢲ(grouping) ࠙ࢳ • ࠺तೣ(ਬࢎࢿ, similarity) - ખ গݒೣ • ౵ۄ޷ఠ • ؘ੉ఠ੄ ౠࢿ ࢶఖ - ݾ੸, ੄ب ղನ • ਬࢎࢿ ӝળ - Cluster੄ ౠ૚ਸ Ѿ੿
10. https://gephi.github.io
11. ௿۞झఠ ࠺૑ب ೟ण(Unsupervised Learning) https://gephi.github.io
12. ؘ੉ఠ ࢠ೒݂ • ആঘഋ / ಞѼ • ׮ࣻ੄ ؘ੉ఠܳ ݻ ѐ੄
 Ӓܛਵ۽ ޘযࢲ ׮ܚ ! • ইਓۄ੉য • औѱ Ӓܛী ࣘೞ૑ ޅ ೞח ؘ੉ఠ http://www.neilson.co.za/k-means-cluster-analysis-in-microsoft-excel/
13. ഝਊ ഑਷ ਽ਊ • ٩۞׬ - FB ঴ҷੋध, ਺ࢿੋध • ஂೱ ࠙ࢳ ߂ ୶ୌ - ಕ੉झ࠘ ஘ҳ୶ୌ, ৲ର • ষ୒ ݆਷ ӝఋ ١١
16. ࠙ࢳೞ۰ݶ • ࠙ೡ ੽Ӕ - K-means • ҅க ੽Ӕ - Dendrogram
17. ௿۞झఠ ࠙ೡ - K-means • N ѐ੄ ؘ੉ఠ • K ѐ੄ ௿۞झఠ۽ ࠙ೡ • ੐੄੄ K
18. ௿۞झఠ ࠙ೡ - K-means 1. K = 2 2. ইޖؘա ੼ 2ѐ 3. N ѐ੄ ؘ੉ఠܳ о੢ о ө਍ ੼җ э਷ Ӓܛਵ۽ ޘ਺ 4. Ӓܛ੄ ಣӐ੼ ҅࢑ 5. ಣӐ੼੉ ߄Շ૑ ঋਸ ٸ ө૑ 3җ 4ܳ ߈ࠂ Ӓې೐ ୹୊: Bishop, C. M. (2006). Pattern Recognition and Machine Learning. Springer.੄ Fig. 9.1
19. ௿۞झఠ ҅க ੽Ӕ • Nѐ੄ ؘ੉ఠ ݽفо ೞաঀ ௿۞झఠ • о੢ ਬࢎೠ ௿۞झఠܳ ೞա੄ ௿۞झఠ۽ ޘ਺ • ௿۞झఠ ղ੄ ؘ੉ఠ৬ о੢ ਬࢎೠ(оө਍) ؘ੉ఠ • ௿۞झఠ ࣻо K ѐо ؼٸ ө૑ ߈ࠂ • ੐੄੄ K
20. ௿۞झఠ ҅க ੽Ӕ http://blog.naver.com/bhserta/60016058486
21. ௿۞झఠ - ؚ٘۽੉٘ http://blog.naver.com/bhserta/60016058486
22. ௿۞झఠ - ؚ٘۽੉٘ http://blog.naver.com/bhserta/60016058486
23. ௿۞झఠ - ݾ੸ http://datawaffle.com/f_DataMining/8959
24. ࠙ࢳೞ۰ݶ • ࠙ೡ ੽Ӕ - K-means • ҅க ੽Ӕ - Dendrogram ! • ੉৻ীب ষ୒ ݆਷ Cluster Model җ Algorithm ઓ੤
25. ࠙ࢳೞ۰ݶ • ࠙ೡ(Partition) ੽Ӕ - k-means, k-medoids, CLARANS • ҅க(Hierarchical) ੽Ӕ - Diana, Agnes, BIRCH, ROCK, CAMELEON • Density ӝ߈ ੽Ӕ - DBSACN, OPTICS, DenClue • Grid ӝ߈ ੽Ӕ - STING, WaveCluster, CLIQUE • Model ӝ߈ ੽Ӕ - EM, SOM, COBWEB • ࠼ب ಁఢ ӝ߈ ੽Ӕ - pCluster • ࢎਊ੗ ӝ߈ ઁড ӝ߈ ੽Ӕ - COD (obstacles), constrained clustering
27. ੿ܻ - ௿۞झఠ ࠙ࢳ • ؀۝੄ ؘ੉ఠܳ ݻ ѐ੄ ౠ૚੸ੋ ޘ਺ਵ۽ ׮ܚ - Grouping, Sampling • ݾ੸ী ٮۄ ࠙ࢳೡ ؘ੉ఠ੄ ౠࢿਸ ࢶఖೞҊ ؘ੉ ఠ ࢎ੉੄ ਬࢎࢿ(഑਷ ࠺ਬࢎࢿ)ਸ ੿੄ • ইਓۄ੉য - औѱ ௿۞झఠী ࣘೞ૑ ঋח ؘ੉ఠ. 
 য়ܨ Ѩ୹੉ա Special Dataܳ ୶୹ী ਽ਊ • ࠺૑ب ೟ण • ௿۞झఠܳ ٜ݅ӝ ਤೠ ߈ࠂ੸ ੘স - ࠻ؘ੉ఠ ౠࢿ
29. gephi ࣗѐ • https://gephi.github.io • http://player.vimeo.com/video/9726202 • Demo
31. ଵҊ 1. Data Journalism, ߓࢿള, http://www.slideshare.net/ hangobi/udis-6-2-data-journalism20140712 2. ؘ੉ఠ݃੉׬ ప௼ץ ੊൤ӝ, imaso, http:// blog.naver.com/bhserta/60016058486 3. ࠻ؘ੉ఠ: ੉ۿ੄ ઙ݈ ইצ ૕ޙ੄ द੘, http:// www.berlinlog.com/?p=1626 4. ਃ્ ڲ׮ח '٩۞׬', ؀୓ Ӓѱ ޤ૑? - ࠶۽ఠ׺֔http:// www.bloter.net/archives/201445