Download presentation
Presentation is loading. Please wait.
Published byKaren Andrews Modified over 9 years ago
1
Clustering Pathways Using Graph Mining Approach Mahmud Shahriar Hossain Monika Akbar Pramodh Pochu Venkata Sesha Sanagavarapu
2
2 Design Pipeline Preprocessor Frequent Subgraph Discovery Graph Objects of Pathways Mined Data Pathway Clustering STKE Dataset NN SearchPathway Relations
3
3 Dataset Properties (size)
4
4
5
5 pf-ipf (tf-idf) TransactionItems bought David Lopez Orange Juice (2), Potato chip (3), Pepsi (1) Robbie Lamb Potato chip (3), Pepsi (3), Beer (1) Jonathan Branden Potato chip (1), Pepsi (1) John Paxton Potato chip (2), Coconut Cookies (2), Pepsi (1) Rafal Angryk Swiss Army Knife (15) Jeannete Radclif Potato chip (2), Coconut Cookies (3) Rocky Ross Orange Juice (2), Coconut Cookies (3) Richard MaClaster Coconut Cookies (3), Beer (1) ………… ……………………………….
6
6 Dataset Properties (pf-ipf)
7
7
8
8 Subgraph Discovery k# of Subgraphs generated Time (sec.) 11,376Existing 25,38041 329,565149 4187,508971 51274,8527518 ------------ min_sup=2% What so novel about pruning edges?
9
9 Subgraph Discovery
10
10 Subgraph Discovery
11
11 Subgraph Discovery
12
12 Subgraph Discovery kNumber of Subgraphs Time Saved (%) Attempts Saved(%) 218699.8398.98 324698.3386.15 430598.5786.38 532398.9586.91 631398.9685.64 727998.8883.25 826398.6778.91 929298.3874.76 1036498.5874.75 1147098.7678.08 1260899.0481.84 1378599.2285.02 1498099.3887.63 15111799.4889.48 16107599.5390.26 1780499.5189.40 1843099.3485.22 1914198.7671.22 20 96.159.19 21175.74-574.47 Overall attempts saved = 89.52% Overall time saved = 99.39%
13
13 Clustering
14
14 Clustering
15
15 Nearest Neighbors Cover Tree and Brute-force method
16
16 Pathway Relations (StoryTelling) Bidirectional Search S p1p1 p2p2 p3p3 T p7p7 p8p8 p9p9
17
17 Pathway Relations (StoryTelling)
18
18 Pathway Relations (StoryTelling)
19
19 Pathway Relations (StoryTelling)
20
20 Questions ???
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.