algorithms presentation
TRANSCRIPT
![Page 1: Algorithms presentation](https://reader036.vdocuments.co/reader036/viewer/2022070519/58ecd7e01a28ab6a118b468f/html5/thumbnails/1.jpg)
How the following Algorithms work
• Clustering
• Collaborative filtering : recommender systems
• Multidimensional scaling
• PCA (Principal Component Analysis)
![Page 2: Algorithms presentation](https://reader036.vdocuments.co/reader036/viewer/2022070519/58ecd7e01a28ab6a118b468f/html5/thumbnails/2.jpg)
Esclusive clusteringAlg.clustering
• Version partitional clustering (Hartigan’s algorithm)
• Version k-mean (random initialization)
![Page 3: Algorithms presentation](https://reader036.vdocuments.co/reader036/viewer/2022070519/58ecd7e01a28ab6a118b468f/html5/thumbnails/3.jpg)
Versione partitional clustering (Hartigan’s alg.)
![Page 4: Algorithms presentation](https://reader036.vdocuments.co/reader036/viewer/2022070519/58ecd7e01a28ab6a118b468f/html5/thumbnails/4.jpg)
K-Means
![Page 5: Algorithms presentation](https://reader036.vdocuments.co/reader036/viewer/2022070519/58ecd7e01a28ab6a118b468f/html5/thumbnails/5.jpg)
Applications.
![Page 6: Algorithms presentation](https://reader036.vdocuments.co/reader036/viewer/2022070519/58ecd7e01a28ab6a118b468f/html5/thumbnails/6.jpg)
Collaborative filtering
• Given a set of users (or more in general objects), and/or preferences, forcast the behavior of the users.
• MovieLens dataset.• Item based CF
![Page 7: Algorithms presentation](https://reader036.vdocuments.co/reader036/viewer/2022070519/58ecd7e01a28ab6a118b468f/html5/thumbnails/7.jpg)
Applications
• Amazon : reccomending articles to users• Facebook : reccomending friends• Netflix : reccomending movies• Google : recomending .. anything
![Page 8: Algorithms presentation](https://reader036.vdocuments.co/reader036/viewer/2022070519/58ecd7e01a28ab6a118b468f/html5/thumbnails/8.jpg)
Multidimensional Scaling
![Page 9: Algorithms presentation](https://reader036.vdocuments.co/reader036/viewer/2022070519/58ecd7e01a28ab6a118b468f/html5/thumbnails/9.jpg)
Multidimensional Scaling 1
![Page 10: Algorithms presentation](https://reader036.vdocuments.co/reader036/viewer/2022070519/58ecd7e01a28ab6a118b468f/html5/thumbnails/10.jpg)
Multidimensional scaling 2-12.5 -12 -11.5 -11 -10.5 -10 -9.5
-16
-15.5
-15
-14.5
-14
-13.5
-13
-12.5
-12
-11.5
Series1
![Page 11: Algorithms presentation](https://reader036.vdocuments.co/reader036/viewer/2022070519/58ecd7e01a28ab6a118b468f/html5/thumbnails/11.jpg)
Multidimensional scaling 3: app
![Page 12: Algorithms presentation](https://reader036.vdocuments.co/reader036/viewer/2022070519/58ecd7e01a28ab6a118b468f/html5/thumbnails/12.jpg)
Dimensionality reduction
• PCA (Principal Component Analysis): eigenvectors decomposition.
• JAMA: Java Matrix library
![Page 13: Algorithms presentation](https://reader036.vdocuments.co/reader036/viewer/2022070519/58ecd7e01a28ab6a118b468f/html5/thumbnails/13.jpg)
Dimensionality reduction2 : app
• Eigenbehaviors: identifying structure in Routine.
• SNA: community affiliation
• PCA + Kmeans = Spectral Clustering: PCA continous sol. <=> discrete sol. k-means clustering
![Page 14: Algorithms presentation](https://reader036.vdocuments.co/reader036/viewer/2022070519/58ecd7e01a28ab6a118b468f/html5/thumbnails/14.jpg)
Dimesionality reduction3: app