The Challenge of Mining Billions of Transactions

Size: px
Start display at page:

Download "The Challenge of Mining Billions of Transactions"

Transcription

1 Faculty of omputer Science The hallenge of Mining illions of Transactions Osmar R. Zaïane International Workshop on lgorithms for Large-Scale Information Processing in Knowledge iscovery Laboratory ata Mining Research Group 008 Size of real data Wal-Mart registers 00 million transactions a week March 00 VIS processes 0 million events a day ecember 007 Yahoo! has 00 million users per month February 008 Why research papers only talk about datasets with about few hundred thousand transactions? Is this realistic scalability? What are the challenges for real large-scale data? in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

2 Roadmap Introduction What are frequent patterns and why are they important? Why is it difficult, it s simple counting after all? Mining xtremely Large atasets: the hallenges Subsets of patterns onstraint pushing Parallelization The Leap and ifold-leap pproaches Improving upon Leap in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Why Studying Frequent Pattern Mining? Frequent pattern mining Foundation for several essential data mining tasks: ssociation rule mining Sequential pattern mining ontrast set mining ssociative classifiers lustering nalysis pplications: basket data analysis, cross-marketing, catalog design, loss-leader analysis, clustering, classification, Web log sequence, N analysis, etc. in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

3 Frequent Itemset Generation null With items we have are = possible candidate itemsets Given d items, there are d possible candidate itemsets in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 priori lgorithm Idea Level-wise algorithm. Generate candidate itemsets of a particular size. Scan the database to see which of them are frequent n itemset is frequent if all its subsets are frequent. Use only these frequent itemsets to generate the set of candidates with size=size+ R. grawal & R. Srikant, VL 99 in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

4 FP-Growth Grow long patterns from short ones using local frequent items abc is a frequent pattern Get all transactions having abc : abc d is a local frequent item in abc abcd is a frequent pattern J. Han, J. Pei, Y. Yin, SIGMO 00 laims to be order of magnitude faster than priori FP-Tree Patterns Recursive conditional trees and FP-Trees in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Illustrating priori Principle Infrequent Items Pruned supersets in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

5 Mining xtremely Large atabases Issues and suggestions? Mining for Frequent patterns is an active research area since the last two decades. Many efficient algorithms exists. Many conferences, and workshops are conducted. So, what is still missing? Mining xtremely large databases is still a problematic issue urrent databases could have hundreds of millions if not billions of transactions Unfortunately we cannot [easily] mine them for frequent patterns in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Mining xtremely Large atabases Issues and suggestions? What to do? ) Investigate finding special patterns (Subsets of the LL patterns) losed patterns Maximal patterns ) Pushing constraints during the mining process ) Parallel implementations of existing sequential algorithms in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

6 Other Frequent Patterns Frequent pattern {a,, a 00 } ( 00 ) + ( 00 ) + + ( ) = 00 - =.7*0 0 frequent sub-patterns! Frequent losed Patterns Frequent Maximal Patterns ll Frequent Patterns ll frequent itemsets losed frequent itemsets Maximal frequent itemsets Maximal frequent itemsets losed frequent itemsets ll frequent itemset in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Frequent losed Patterns For frequent itemset X, if there exists no item y such that every transaction containing X also contains y, then X is a frequent closed pattern In other words, frequent itemset X is closed if there is no item y, not already in X, that always accompanies X in all transactions where X occurs. oncise representation of frequent patterns. an generate all frequent patterns with their support from frequent closed ones. Reduce number of patterns and rules N. Pasquier et al. In IT 99 in Knowledge iscovery PK 008 {abcd} {abc} {bd} Transactions Support = a b c d ab ac bc bd abc Frequent itemsets b bd abc Frequent losed itemsets Osaka, Japan May 0, 008 O.R. Zaïane 008

7 Frequent Maximal Patterns Frequent itemset X is maximal if there is no other frequent itemset Y that is superset of X. In other words, there is no other frequent pattern that would include a maximal pattern. More concise representation of frequent patterns but the information about supports is lost. an generate all frequent patterns from frequent maximal ones but without their respective support. R. ayardo. In SIGMO 98 in Knowledge iscovery PK 008 {abcd} {abc} {bd} Transactions Support = a b c d ab ac bc bd abc Frequent itemsets bd abc Frequent Maximal itemsets Osaka, Japan May 0, 008 O.R. Zaïane 008 TI Items Maximal vs. losed Itemsets null Transaction Ids Not supported by any transactions in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

8 Maximal vs. losed Itemsets null losed but not maximal losed and maximal Frequent Pattern order Minimum support = in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 onstraint-based frequent patterns Finding all the patterns in a database autonomously? unrealistic! The patterns could be too many but not focused! ata mining should be an interactive process User directs what to be mined using a data mining query language (or a graphical user interface) onstraint-based mining User flexibility: provides constraints on what to be mined System optimization: explores such constraints for efficient mining constraint-based mining in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

9 nti-monotonicity in onstraint-ased Mining nti-monotonicity When an intemset S violates the constraint, so does any of its superset sum(s.price) v is anti-monotone sum(s.price) v is not anti-monotone xample. : range(s.profit) is antimonotone Itemset ab violates So does every superset of ab TI T (min_sup=) Item a b c d e f Transaction a, b, c, d, f b, c, d, f, g, h a, c, d, e, f c, e, f, g Profit g h 0-0 in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Monotonicity in onstraint-ased Mining T (min_sup=) Monotonicity When an intemset S satisfies the constraint, so does any of its superset sum(s.price) v is monotone TI Transaction a, b, c, d, f b, c, d, f, g, h a, c, d, e, f c, e, f, g Item Profit min(s.price) v is monotone xample. : range(s.profit) Itemset ab satisfies So does every superset of ab a b c d e f g h in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

10 The onstraints FIM lgorithms (State of the rt) FI M J. Pie and J. Han. SIGK 000 ual Miner. ucil, J. Gherke,. Kiefer and W. White, SIGK 0 xaminer F. onchi, F. Giannotti,. Mazzanti, and. Pedreschi. IM 00 FP-onsai F. onchi and. Goethals. PK 00 ifoldleap M. l-hajj, P. Nalos and O. Zaiane, IM 0 in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Parallel Frequent itemset RPLIT candidate sets lgorithms andidate Set andidate Set andidate Set Replication lgorithms Partitioning lgorithms Hybrid lgorithms Replicate andidate Sets Partition the databases Partitioned database ount istribution algorithm (grawal R. et al, 99) Parallel Partition algorithm (shok Savasere et al, 99) Fast istributed algorithm (avid heung et al, 99) Parallel ata Mining algorithm (J.S. Park et al, 99) in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

11 Parallel Frequent itemset lgorithms P. andidate Set andidate sets Partitioned P. andidate Set P. andidate Set Replication lgorithms Partitioning lgorithms Hybrid lgorithms -Partition the andidate Set -Need to Scan the whole data database ata istribution algorithm (grawal R. et al, 99) Intelligent ata istribution algorithm (-H Han et all, 997) Non Partitioned apriori, Simple partitioned, Hash Partitioned apriori and HP_L algorithms (Takahiko et al, 99) in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Parallel Frequent itemset lgorithms P. andidate Set andidate sets Partitioned P. andidate Set P. andidate Set Replication lgorithms Partitioning lgorithms Hybrid lgorithms -ombine both ideas of replication and partition Partitioned database Hybrid istribution algorithm (-H Han et all, 997) in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

12 How to mine extremely large databases? What can help us in mining large databases ) Investigate finding special patterns losed patterns Maximal patterns ) Pushing constraints ) Parallel implementations of existing sequential algorithms Still Mining extremely large databases is not feasible in practice in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 So, Where is the problem? Open Problems exists such as: How to mine extremely large databases? What can help us in mining large databases ) Investigate finding special patterns losed patterns Maximal patterns ) Pushing constraints ) Parallel implementations of existing sequential algorithms.observations on Superfluous Processing.Superfluous Search of the Lattice.Huge Memory structures used.parallel implementations are mainly priori based and parallelization of existing algorithms in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

13 Transaction Layout Horizontal Layout ach transaction is recorded as a list of items Transaction I Items G H M F N N O P Q R G 7 H I G 8 L F K 9 F M N O 0 F P J R H I K L M G O F P Q J F I J 7 K F 8 L andidacy generation can be removed [or reduced] (FP-Growth) Superfluous Processing in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Transaction Layout Vertical Layout Tid-list is kept for each item Transaction I Items G H M F N N O P Q R G 7 H I G 8 L F K 9 F M N O 0 F P J R H I K L M G O F P Q J F I J 7 K F 8 L Items Transaction I F G 7 H 7 I 7 J 0 K 8 7 L 8 8 M 9 N 9 O 9 P 9 Q R 0 Minimize Superfluous Processing andidacy generation is required in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

14 Transaction Layout itmap Layout Matrix : Rows represent transactions olumns represent item If item exists in transaction then cell value = else cell value = 0 T# Items T G T H T M T F N T N O P T Q R G T7 H I G T8 L F K T9 F M N O T0 F P J R T H I T K L T M G O T F P Q J T F I T J T7 K F T8 L Transaction I items T# F G H I J K L M N O P Q R T T T T T T T T T T T T T T T T T T Similar to horizontal layout. Suitable for datasets with small dimensionality in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Inverted Matrix Layout Interactive mining hanging the support level means expensive steps (whole process is redone) valuation and Presentation Knowledge Selection and Transformation ata warehouse ata Mining Patterns atabases in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

15 Why the Matrix Layout Repetitive tasks, (I/O) readings (Superfluous Processing) T# Items T# Items T G T G T H T G T M T G T F N T F G Support > T N O P T G G G T Q R G T G G G T7 H I G T7 G G G T8 L F K Frequent -itemsets {,,,,, F} T8 G F G T9 F M N O Non frequent items {G, H, I, J, K, L, M, N, O, P, Q, R} T9 F G G G T0 F P J R T0 F G G G T H I T G G T K L T M G O T F P Q J T G G T G G G T F G G G T F I T F G T J T G T7 K F T7 G F T8 L T8 G in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Why the Matrix Layout Repetitive tasks, (I/O) readings (Superfluous Processing) T# T Items G T H T M T F N T N O P T Q R G T7 H I G T8 L F K T9 F M N O T0 F P J R T H I T K L T M G O T F P Q J T F I T J T7 K F T8 L Support > 9 Frequent -itemsets {,, } Non frequent items {,, F, G, H, I, J, K, L, M, N, O, P, Q, R} T# T Items G G T G G G T G G G T G G G T G G G T G G G T7 G G G T8 G G G G T9 G G G G T0 G G G G T G G G T G G G G T G G G G T G G G G T G G G G T G G G T7 G G G T8 G G in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

16 Why the Matrix Layout Loc Index R Q P O N M 7 L 8 K 9 J 0 I H G F Transactional rray Transaction I Items G H M F N N O P Q R G 7 H I G 8 L F K 9 F M N O 0 F P J R H I K L M G O F P Q J F I J 7 K F 8 L in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Why the Matrix Layout <G,,,,> Loc Index R Q P O N M 7 L 8 K 9 J 0 I H G F Transactional rray T# Items T G T H T M T F N T N O P T Q R G T7 H I G T8 L F K T9 F M N O T0 F P J R T H I T K L T M G O T F P Q J T F I T J T7 K F T8 L in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

17 Why the Matrix Layout <G,,,,> <H,,,,> Loc Index Transactional rray R Q P O N M 7 L 8 K 9 J 0 I H (,) G (,) F 7 8 (,) 9 (,) (,) 0 (7,) (7,) 7 0 (8,) (, ) 8 (, ) T# Items T G T H T M T F N T N O P T Q R G T7 H I G T8 L F K T9 F M N O T0 F P J R T H I T K L T M G O T F P Q J T F I T J T7 K F T8 L in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Why the Matrix Layout Loc Index Transactional rray R (,) (,) Q (,) (,) P (,) (9,) (9,) O (,) (,) (,) N (,) (7,) (,) M (,) (,) (,) 7 L (8,) (8,) (,9) 8 K (,) (,) (,7) 9 J (,) (,) (,7) 0 I (,) (,) (,) H (,) (,),) G (,) (,) (,) (,) F 7 (,) (,) (8,7) (,) (,8) (,) (,8) 8 (,) (,) (,) (7,) (,) (,7) (,8) (,9) 9 (,) (,) (7,) (7,) (7,7) (,7) (7,8) (7,9) (,0) 0 (7,) (7,) (8,) (8,) (8,) (, ) (, ) (, ) (8,0) (7,0) 7 0 (8,) (, ) (8,) (8,) (, ) (8,8) (, ) (, ) (8,9) (8,) 8 (, ) (, ) (, ) (, ) (, ) (, ) (, ) (, ) (, ) (, ) (, ) in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

18 Why the Matrix Layout Inverted Matrix Layout Support > Loc Index Transactional rray R (,) (,) Q (,) (,) P (,) (9,) (9,) O (,) (,) (,) N (,) (7,) (,) M (,) (,) (,) 7 L (8,) (8,) (,9) 8 K (,) (,) (,7) 9 J (,) (,) (,7) 0 I (,) (,) (,) H (,) (,),) G (,) (,) (,) (,) F 7 (,) (,) (8,7) (,) (,8) (,) (,8) 8 (,) (,) (,) (7,) (,) (,7) (,8) (,9) 9 (,) (,) (7,) (7,) (7,7) (,7) (7,8) (7,9) (,0) 0 (7,) (7,) (8,) (8,) (8,) (, ) (, ) (, ) (8,0) (7,0) 7 0 (8,) (, ) (8,) (8,) (, ) (8,8) (, ) (, ) (8,9) (8,) 8 (, ) (, ) (, ) (, ) (, ) (, ) (, ) (, ) (, ) (, ) (, ) order Support in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Why the Matrix Layout Loc Index Transactional rray F 7 (,) (,) (8,7) (,) (,8) (,) (,8) 8 (,) (,) (,) (7,) (,) (,7) (,8) (,9) 9 (,) (,) (7,) (7,) (7,7) (,7) (7,8) (7,9) (,0) 0 (7,) (7,) (8,) (8,) (8,) (, ) (, ) (, ) (8,0) (7,0) 7 0 (8,) (, ) (8,) (8,) (, ) (8,8) (, ) (, ) (8,9) (8,) 8 (, ) (, ) (, ) (, ) (, ) (, ) (, ) (, ) (, ) (, ) (, ) in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

19 Why the Matrix Layout Loc Index Transactional rray F 7 (,) (,) (8,7) (,) (,8) (,) (,8) 8 (,) (,) (,) (7,) (,) (,7) (,8) (,9) 9 (,) (,) (7,) (7,) (7,7) (,7) (7,8) (7,9) (,0) 0 (7,) (7,) (8,) (8,) (8,) (, ) (, ) (, ) (8,0) (7,0) 7 0 (8,) (, ) (8,) (8,) (, ) (8,8) (, ) (, ) (8,9) (8,) 8 (, ) (, ) (, ) (, ) (, ) (, ) (, ) (, ) (, ) (, ) (, ) T# T Items T T T F T T T7 T8 F T9 F T0 F T T T T F T F T T7 F T8 in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Mining the Pattern Lattice (Traversal pproaches) ottom-up It uses current items at level k to generate items of level k+ (many database scans) Top-own It uses current items at level k+ to generate items of level k (favored when mining long frequent patterns) Hybrid null readth epth Hybrid approach It mines using both direction at the same time in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

20 Leap Traversal xample null Steps ottom x TI Items x Itemset is candidate if it is marked or if it is a subset of more than one infrequent marked superset How to find the x x Support of an itemset. Full scan of the database OR. Intelligent techniques: Support of itemset = Top Summation of the supports of its supersets of marked patterns Frequent Path bases in Knowledge iscovery PK candidates to check frequent patterns without checking Osaka, Japan May 0, 008 O.R. Zaïane 008 How can we generate Frequent Path ases? OFI trees Header less Frequent Pattern tree in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

21 OFI algorithm big picture OFI I/O scans reduced candidacy generation Small memory footprint FP-Tree OFI- trees Patterns in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 How to Find Frequent Path ases? Headless Frequent Pattern tree example: Transactional database mined with support >= in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

22 How to Find Frequent Path ases? Headless Frequent Pattern tree example: Find Frequent Items in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 How to Find Frequent Path ases? Headless Frequent Pattern tree example: Remove none frequent items in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

23 How to Find Frequent Path ases? Headless Frequent Pattern tree example: Sort Frequent items in escending order based on their support in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 How to Find Frequent Path ases? Headless Frequent Pattern tree example:,,0 NULL,,0,,0,,0 in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

24 How to Find Frequent Path ases? Headless Frequent Pattern tree example:,,0 NULL,,0,,0,,0,,0 in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 How to Find Frequent Path ases? Headless Frequent Pattern tree example:,,0 NULL,,0,,0,,0,,0,,0,,0,,0 in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

25 How to Find Frequent Path ases? Headless Frequent Pattern tree example: Scan From Leaf to root to generate FP NULL Support ranch = Support ranch support,7,0,9,0,,0 Support Participation,,0,,0,,0,,0,,0,,0,,0,,0,,0 in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 How to Find Frequent Path ases? Headless Frequent Pattern tree example: Occurs times in Occurs time in Frequency counting What is the support of,, Occurs times in Total Support = Frequent Path ases Support ranch in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

26 How to find the maximal patterns? y all applying all possible intersection among the FP (asic pproach) NUL L in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Pruning strategies () Pruning Strategies For all X, and Y in FPs ordered lexicographically, if X Y is frequent then there is no need to intersect any other elements that have X Y NUL., i.e, all children of X Y can be pruned L in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

27 Pruning strategies () Pruning Strategies For all X, and Y in FPs ordered lexicographically, if X Y is frequent then there is no need to intersect any other elements that have X Y NUL., i.e, all children of X Y can be pruned L in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Pruning strategies () Pruning Strategies For all X, and Y in FPs ordered lexicographically, if X Y is frequent then there is no need to intersect any other elements that have X Y NUL., i.e, all children of X Y can be pruned L in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

28 Pruning strategies () For all X, Y, and W in FPs ordered lexicographically, if X Y = X W and Y << W (i.e Y is left of Win the lexicographic tree) then there is no need to explore any children of X Y. Pruning Strategies NUL L ssume = in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Pruning strategies () For all X, Y, and W in FPs ordered lexicographically, if X Y = X W and Y << W (i.e Y is left of Win the lexicographic tree) then there is no need to explore any children of X Y. Pruning Strategies NUL L ssume = p in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

29 Pruning strategies () For all X, Y, and W in FPs ordered lexicographically, if X Y = X W and Y << W (i.e Y is left of Win the lexicographic tree) then there is no need to explore any children of X Y. Pruning Strategies NUL L ssume = in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Pruning strategies () For all X, Y,Z in FPs ordered lexicographically, if X Y is subset of X Z then we can ignore the sub-tree X Y Z. Pruning Strategies NUL L ssume is subset of in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

30 Pruning strategies () For all X, Y,Z in FPs ordered lexicographically, if X Y is subset of X Z then we can ignore the sub-tree X Y Z. Pruning Strategies NUL L ssume is subset of in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Pruning strategies () For all X, Y,Z in FPs ordered lexicographically, if X Y is subset of X Z then we can ignore the sub-tree X Y Z. Pruning Strategies NUL L ssume is subset of in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

31 Pruning strategies () For all X, Y,Z in FPs, if X Y is superset of X Z then we can ignore the sub-tree of X Z as long X Z is not Frequent Pruning Strategies NUL L ssume is a superset of in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Pruning strategies () For all X, Y,Z in FPs, if X Y is superset of X Z then we can ignore the sub-tree of X Z as long X Z is not Frequent Pruning Strategies NUL L ssume is a superset of in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

32 Pruning strategies () For all X, Y,Z in FPs, if X Y is superset of X Z then we can ignore the sub-tree of X Z as long X Z is not Frequent Pruning Strategies NUL L ssume is a superset of in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Pruning strategies () For all X, Y,Z in FPs, if X Y is superset of X Z then we can ignore the sub-tree of X Z as long X Z is not Frequent Pruning Strategies NUL L ssume is a superset of in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

33 The Set of Maximal Patterns t this stage we have the set of Maximal Patterns with the set of Frequent Path ases. We can generate all subsets of this maximals with their support using the FP branch supports In our implementation we used bitmaps for itemset presentations: N Operation is used for the intersections and for subset checking Using The set of Maximal Patterns with the FP we can generate the set of LL or LOS Patterns lgorithms are either called OFI-Leap or HFP-Leap in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 xperiments (Samples on real data) High Support Low Support in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

34 Mining relatively small synthetic datasets mining for LL patterns in small synthetic dataset mining for LOS patterns in small synthetic dataset Time in seconds OFI-LL FP-Growth MFI Time in seconds OFI-LOS FPLOS MFI K 0K 00K 0K 0 0K 0K 00K 0K transaction size transaction size mining for MXIML patterns in synthetic dataset Time in seconds OFI-MX FPMX MFI 0 0K 0K 00K 0K transaction size in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 xperiments (Memory Usage) Support = %, % of Memory usage. HFP-Leap MFI % OFI-Leap % % FPMX 8% GNMX 7% OFI-Leap HFP-Leap GNMX FPMX MFI Support = %, % of Memory usage. Mining Retail ataset OFI-Leap MFI % % FPMX 8% GNMX 7% HFP-Leap % OFI-Leap HFP-Leap GNMX FPMX MFI in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

35 xperiments (xtremely Large datasets) in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Leap with onstraints (ifoldleap) Push both types of constraints (Monotone and nti-monotone) in duality. uring the mining process Two Strategies are being adopted:. Push onstraints to prune. void onstraints checking in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

36 ifold Leap (Leap with constraints) F F F F F G H I J K L J K L F G H I Intersecting FP Frequent-Path-ases F F,,,, F F F F F F F F F F F F F F F F F F F F F F F F F F F F F Head (H): Is a frequent path base or any subset generated from the intersection of frequent path bases Tail (T): Is the itemset generated from intersecting all remaining frequent path bases not used in the intersection of (H) in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 ifold Leap (Leap with constraints) Theorem : If an intersection of frequent path bases (H) fails Q(), it can be discarded, and there is no need to evaluate further intersections with (H). F Fails Q() Monotone onstraints F F F F F F ll subsets will also fails Q() F F F F F F F F F F F F F TOP OWN TSTING in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

37 ifold Leap (Leap with constraints) Theorem : If an intersection of frequent path bases (H) fails Q(), it can be discarded, and there is no need to evaluate further intersections with (H). F F F F F Fails Q() ll subsets will also fails Q() Prune H() with all its subsets F F F TOP OWN TSTING in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 ifold Leap (Leap with constraints) Theorem : If an intersection of frequent path bases (H) passes P(), it is a candidate P-maximal, and there is no need to evaluate further intersections with (H). F Passes P() nti-monotone onstraints F F F F F F ll subsets will also Pass P() F F F F F F F F F F F F F TOP OWN TSTING in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

38 ifold Leap (Leap with constraints) Theorem : If an intersection of frequent path bases (H) passes P(), it is a candidate P-maximal, and there is no need to evaluate further intersections with (H). F F F F F F F F F F F F F F F F Passes P() ll subsets will also Pass P() No need to test against P() for all subsets of (H) F F F F Reduce evaluations numbers TOP OWN TSTING in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 ifold Leap (Leap with constraints) Theorem : If a node s (H) (T) fails P(), the (H) node can be discarded, and there is no need to evaluate further intersections with (H). F F F F F Fails P() ll Superset will also fails P() F F F OTTOM UP TSTING in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

39 ifold Leap (Leap with constraints) Theorem : If a node s (H) (T) fails P(), the (H) node can be discarded, and there is no need to evaluate further intersections with (H). Fails P() ll Superset will also fails P() Prune H() with all its subsets OTTOM UP TSTING in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 ifold Leap (Leap with constraints) Theomrem :If a node s (H) (T) passes Q(), Q() is guaranteed to pass for any itemset resulting from the intersection of a subset of the frequent path bases used to generate (H) plus the remaining frequent path bases yet to be intersected with (H). Q() does not need to be checked in these cases. F F F F F Pass Q() ll Superset will also pass Q() F F F OTTOM UP TSTING in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

40 ifold Leap (Leap with constraints) Theomrem :If a node s (H) (T) passes Q(), Q() is guaranteed to pass for any itemset resulting from the intersection of a subset of the frequent path bases used to generate (H) plus the remaining frequent path bases yet to be intersected with (H). Q() does not need to be checked in these cases. F F F F F F F F Pass Q() ll Superset will also fails Q() No need to test against Q() for all supersets of (H) Reduce evaluations numbers OTTOM UP TSTING in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 i-irectional Pushing of constraints Theorems applied at the (H) F F F F F F F F F F F F F F F F Theorem : To prune subsets Theorem : To reduce constraints evaluations Theorems applied at the (H T) Theorem : To prune supersets Theorem : To reduce constraints evaluations F F F F in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

41 Results of ifold Leap Set of maximal patterns that satisfies both P() and Q() (P-Maximal) ll subsets of these patterns are guaranteed to satisfy P() The subsets are only testes against Q() The final sets are the set of frequent patterns that satisfies both P() and Q() in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Performance valuation ifoldleap (Q only) ualminer (Q only) ifoldleap (P only) ualminer (P only) ifoldleap (P & Q) ualminer (P & Q) retail dataset 00 Pushing P(), Q() and P() N Q() on the retail datasets Time in seconds Support 0.7% 0.% 0.08% ifoldleap (Q only) ualminer (Q only) ifoldleap (P only) ualminer (P only) ifoldleap (P & Q) ualminer (P & Q) retail dataset Time in seconds Support 0.7% 0.% 0.08% in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

42 Performance valuation Scalability tests ifoldleap (P & Q) ualminer (P & Q) Time in seconds transaction size 0K 00K 0K 00K in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Parallel Leap in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

43 Parallel Leap xample: Processors. in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Parallel Leap xample: Processors. Trees istributions. irect. Round Robin. First Last in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

44 Parallel ifold-leap xample: in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Parallel ifold-leap xample: Trees istributions. irect. Round Robin. First Last in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

45 Parallel xperiments Our experiments are conducted using a cluster made of boxes ach box has a linux..8 dual processor.ghz M,. G RM. Nodes are connected using Fast ethernet and Myrinet 000 network in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 xperiments Test Scalability: ifferent transactions sizes (p=) in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

46 xperiments Test Scalability: ifferent number of processors (TS = 00Millions) in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 xperiments Test Speedup: ifferent support values (TS = 00Millions) in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

47 xperiments ffect of Leap node distribution in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 onclusion and Future work framework of sequential and parallel algorithms can be implemented to mine extremely large databases by applying two strategies which are: y using a novel traversal searching for frequent patterns while using minimal memory requirements. y designing algorithms specially made for parallel execution rather than parallelizing an algorithm designed for a sequential execution. in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

48 onclusion and Future work Open Questions are still exist in this research area What algorithm is considered the best? How to determine database characteristics? What patterns to generate? in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Heterogeneous omputer lusters omputer clusters evolve in time by addition and suppression of nodes lusters are bound to become heterogeneous omputer clusters are now distributed geographically (grids) How to manage parallel mining and dynamic load balancing on heterogeneous clusters? How to manage parallel mining when the pipeline becomes the bottleneck and is inconsistent. in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

49 symmetric Parallel ata Mining Frequent itemset mining: Many algorithms exist No real winner epends on dataset What determines the best algorithm to use given a dataset? partition a b c d n Scenario ll nodes use the same algorithm Scenario Nodes use different algorithms etermine new data partitioning strategies. reate new communication protocols. evise new load balancing techniques. Invent new parallel algorithms. in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008 Thank you UNIVRSITY OF LRT Osmar R. Zaïane, Ph.. ssociate Professor thabasca Hall dmonton, lberta anada TG 8 Telephone: Office + (780) 9 80 Fax + (780) mail: zaiane@cs.ualberta.ca Questions? in Knowledge iscovery PK 008 Osaka, Japan May 0, 008 O.R. Zaïane 008

Frequent Pattern Mining. Toon Calders University of Antwerp

Frequent Pattern Mining. Toon Calders University of Antwerp Frequent Pattern Mining Toon alders University of ntwerp Summary Frequent Itemset Mining lgorithms onstraint ased Mining ondensed Representations Frequent Itemset Mining Market-asket nalysis transaction

More information

Association Rules. Fundamentals

Association Rules. Fundamentals Politecnico di Torino Politecnico di Torino 1 Association rules Objective extraction of frequent correlations or pattern from a transactional database Tickets at a supermarket counter Association rule

More information

D B M G Data Base and Data Mining Group of Politecnico di Torino

D B M G Data Base and Data Mining Group of Politecnico di Torino Data Base and Data Mining Group of Politecnico di Torino Politecnico di Torino Association rules Objective extraction of frequent correlations or pattern from a transactional database Tickets at a supermarket

More information

D B M G. Association Rules. Fundamentals. Fundamentals. Association rules. Association rule mining. Definitions. Rule quality metrics: example

D B M G. Association Rules. Fundamentals. Fundamentals. Association rules. Association rule mining. Definitions. Rule quality metrics: example Association rules Data Base and Data Mining Group of Politecnico di Torino Politecnico di Torino Objective extraction of frequent correlations or pattern from a transactional database Tickets at a supermarket

More information

D B M G. Association Rules. Fundamentals. Fundamentals. Elena Baralis, Silvia Chiusano. Politecnico di Torino 1. Definitions.

D B M G. Association Rules. Fundamentals. Fundamentals. Elena Baralis, Silvia Chiusano. Politecnico di Torino 1. Definitions. Definitions Data Base and Data Mining Group of Politecnico di Torino Politecnico di Torino Itemset is a set including one or more items Example: {Beer, Diapers} k-itemset is an itemset that contains k

More information

CS 484 Data Mining. Association Rule Mining 2

CS 484 Data Mining. Association Rule Mining 2 CS 484 Data Mining Association Rule Mining 2 Review: Reducing Number of Candidates Apriori principle: If an itemset is frequent, then all of its subsets must also be frequent Apriori principle holds due

More information

Data Mining: Concepts and Techniques. (3 rd ed.) Chapter 6

Data Mining: Concepts and Techniques. (3 rd ed.) Chapter 6 Data Mining: Concepts and Techniques (3 rd ed.) Chapter 6 Jiawei Han, Micheline Kamber, and Jian Pei University of Illinois at Urbana-Champaign & Simon Fraser University 2013 Han, Kamber & Pei. All rights

More information

Data Mining Concepts & Techniques

Data Mining Concepts & Techniques Data Mining Concepts & Techniques Lecture No. 04 Association Analysis Naeem Ahmed Email: naeemmahoto@gmail.com Department of Software Engineering Mehran Univeristy of Engineering and Technology Jamshoro

More information

DATA MINING LECTURE 3. Frequent Itemsets Association Rules

DATA MINING LECTURE 3. Frequent Itemsets Association Rules DATA MINING LECTURE 3 Frequent Itemsets Association Rules This is how it all started Rakesh Agrawal, Tomasz Imielinski, Arun N. Swami: Mining Association Rules between Sets of Items in Large Databases.

More information

Course Content. Association Rules Outline. Chapter 6 Objectives. Chapter 6: Mining Association Rules. Dr. Osmar R. Zaïane. University of Alberta 4

Course Content. Association Rules Outline. Chapter 6 Objectives. Chapter 6: Mining Association Rules. Dr. Osmar R. Zaïane. University of Alberta 4 Principles of Knowledge Discovery in Data Fall 2004 Chapter 6: Mining Association Rules Dr. Osmar R. Zaïane University of Alberta Course Content Introduction to Data Mining Data warehousing and OLAP Data

More information

Lecture Notes for Chapter 6. Introduction to Data Mining

Lecture Notes for Chapter 6. Introduction to Data Mining Data Mining Association Analysis: Basic Concepts and Algorithms Lecture Notes for Chapter 6 Introduction to Data Mining by Tan, Steinbach, Kumar Tan,Steinbach, Kumar Introduction to Data Mining 4/18/2004

More information

CS 584 Data Mining. Association Rule Mining 2

CS 584 Data Mining. Association Rule Mining 2 CS 584 Data Mining Association Rule Mining 2 Recall from last time: Frequent Itemset Generation Strategies Reduce the number of candidates (M) Complete search: M=2 d Use pruning techniques to reduce M

More information

ASSOCIATION ANALYSIS FREQUENT ITEMSETS MINING. Alexandre Termier, LIG

ASSOCIATION ANALYSIS FREQUENT ITEMSETS MINING. Alexandre Termier, LIG ASSOCIATION ANALYSIS FREQUENT ITEMSETS MINING, LIG M2 SIF DMV course 207/208 Market basket analysis Analyse supermarket s transaction data Transaction = «market basket» of a customer Find which items are

More information

Chapters 6 & 7, Frequent Pattern Mining

Chapters 6 & 7, Frequent Pattern Mining CSI 4352, Introduction to Data Mining Chapters 6 & 7, Frequent Pattern Mining Young-Rae Cho Associate Professor Department of Computer Science Baylor University CSI 4352, Introduction to Data Mining Chapters

More information

Association Rule. Lecturer: Dr. Bo Yuan. LOGO

Association Rule. Lecturer: Dr. Bo Yuan. LOGO Association Rule Lecturer: Dr. Bo Yuan LOGO E-mail: yuanb@sz.tsinghua.edu.cn Overview Frequent Itemsets Association Rules Sequential Patterns 2 A Real Example 3 Market-Based Problems Finding associations

More information

Data Mining and Analysis: Fundamental Concepts and Algorithms

Data Mining and Analysis: Fundamental Concepts and Algorithms Data Mining and Analysis: Fundamental Concepts and Algorithms dataminingbook.info Mohammed J. Zaki 1 Wagner Meira Jr. 2 1 Department of Computer Science Rensselaer Polytechnic Institute, Troy, NY, USA

More information

COMP 5331: Knowledge Discovery and Data Mining

COMP 5331: Knowledge Discovery and Data Mining COMP 5331: Knowledge Discovery and Data Mining Acknowledgement: Slides modified by Dr. Lei Chen based on the slides provided by Jiawei Han, Micheline Kamber, and Jian Pei And slides provide by Raymond

More information

732A61/TDDD41 Data Mining - Clustering and Association Analysis

732A61/TDDD41 Data Mining - Clustering and Association Analysis 732A61/TDDD41 Data Mining - Clustering and Association Analysis Lecture 6: Association Analysis I Jose M. Peña IDA, Linköping University, Sweden 1/14 Outline Content Association Rules Frequent Itemsets

More information

Chapter 6. Frequent Pattern Mining: Concepts and Apriori. Meng Jiang CSE 40647/60647 Data Science Fall 2017 Introduction to Data Mining

Chapter 6. Frequent Pattern Mining: Concepts and Apriori. Meng Jiang CSE 40647/60647 Data Science Fall 2017 Introduction to Data Mining Chapter 6. Frequent Pattern Mining: Concepts and Apriori Meng Jiang CSE 40647/60647 Data Science Fall 2017 Introduction to Data Mining Pattern Discovery: Definition What are patterns? Patterns: A set of

More information

FP-growth and PrefixSpan

FP-growth and PrefixSpan FP-growth and PrefixSpan n Challenges of Frequent Pattern Mining n Improving Apriori n Fp-growth n Fp-tree n Mining frequent patterns with FP-tree n PrefixSpan Challenges of Frequent Pattern Mining n Challenges

More information

A New Concise and Lossless Representation of Frequent Itemsets Using Generators and A Positive Border

A New Concise and Lossless Representation of Frequent Itemsets Using Generators and A Positive Border A New Concise and Lossless Representation of Frequent Itemsets Using Generators and A Positive Border Guimei Liu a,b Jinyan Li a Limsoon Wong b a Institute for Infocomm Research, Singapore b School of

More information

CSE 5243 INTRO. TO DATA MINING

CSE 5243 INTRO. TO DATA MINING CSE 5243 INTRO. TO DATA MINING Mining Frequent Patterns and Associations: Basic Concepts (Chapter 6) Huan Sun, CSE@The Ohio State University Slides adapted from Prof. Jiawei Han @UIUC, Prof. Srinivasan

More information

DATA MINING - 1DL360

DATA MINING - 1DL360 DATA MINING - 1DL36 Fall 212" An introductory class in data mining http://www.it.uu.se/edu/course/homepage/infoutv/ht12 Kjell Orsborn Uppsala Database Laboratory Department of Information Technology, Uppsala

More information

DATA MINING LECTURE 4. Frequent Itemsets, Association Rules Evaluation Alternative Algorithms

DATA MINING LECTURE 4. Frequent Itemsets, Association Rules Evaluation Alternative Algorithms DATA MINING LECTURE 4 Frequent Itemsets, Association Rules Evaluation Alternative Algorithms RECAP Mining Frequent Itemsets Itemset A collection of one or more items Example: {Milk, Bread, Diaper} k-itemset

More information

Encyclopedia of Machine Learning Chapter Number Book CopyRight - Year 2010 Frequent Pattern. Given Name Hannu Family Name Toivonen

Encyclopedia of Machine Learning Chapter Number Book CopyRight - Year 2010 Frequent Pattern. Given Name Hannu Family Name Toivonen Book Title Encyclopedia of Machine Learning Chapter Number 00403 Book CopyRight - Year 2010 Title Frequent Pattern Author Particle Given Name Hannu Family Name Toivonen Suffix Email hannu.toivonen@cs.helsinki.fi

More information

Unit II Association Rules

Unit II Association Rules Unit II Association Rules Basic Concepts Frequent Pattern Analysis Frequent pattern: a pattern (a set of items, subsequences, substructures, etc.) that occurs frequently in a data set Frequent Itemset

More information

Association Rules. Acknowledgements. Some parts of these slides are modified from. n C. Clifton & W. Aref, Purdue University

Association Rules. Acknowledgements. Some parts of these slides are modified from. n C. Clifton & W. Aref, Purdue University Association Rules CS 5331 by Rattikorn Hewett Texas Tech University 1 Acknowledgements Some parts of these slides are modified from n C. Clifton & W. Aref, Purdue University 2 1 Outline n Association Rule

More information

Association Analysis Part 2. FP Growth (Pei et al 2000)

Association Analysis Part 2. FP Growth (Pei et al 2000) Association Analysis art 2 Sanjay Ranka rofessor Computer and Information Science and Engineering University of Florida F Growth ei et al 2 Use a compressed representation of the database using an F-tree

More information

Association Rules Information Retrieval and Data Mining. Prof. Matteo Matteucci

Association Rules Information Retrieval and Data Mining. Prof. Matteo Matteucci Association Rules Information Retrieval and Data Mining Prof. Matteo Matteucci Learning Unsupervised Rules!?! 2 Market-Basket Transactions 3 Bread Peanuts Milk Fruit Jam Bread Jam Soda Chips Milk Fruit

More information

DATA MINING - 1DL360

DATA MINING - 1DL360 DATA MINING - DL360 Fall 200 An introductory class in data mining http://www.it.uu.se/edu/course/homepage/infoutv/ht0 Kjell Orsborn Uppsala Database Laboratory Department of Information Technology, Uppsala

More information

Positive Borders or Negative Borders: How to Make Lossless Generator Based Representations Concise

Positive Borders or Negative Borders: How to Make Lossless Generator Based Representations Concise Positive Borders or Negative Borders: How to Make Lossless Generator Based Representations Concise Guimei Liu 1,2 Jinyan Li 1 Limsoon Wong 2 Wynne Hsu 2 1 Institute for Infocomm Research, Singapore 2 School

More information

Frequent Itemsets and Association Rule Mining. Vinay Setty Slides credit:

Frequent Itemsets and Association Rule Mining. Vinay Setty Slides credit: Frequent Itemsets and Association Rule Mining Vinay Setty vinay.j.setty@uis.no Slides credit: http://www.mmds.org/ Association Rule Discovery Supermarket shelf management Market-basket model: Goal: Identify

More information

Data Analytics Beyond OLAP. Prof. Yanlei Diao

Data Analytics Beyond OLAP. Prof. Yanlei Diao Data Analytics Beyond OLAP Prof. Yanlei Diao OPERATIONAL DBs DB 1 DB 2 DB 3 EXTRACT TRANSFORM LOAD (ETL) METADATA STORE DATA WAREHOUSE SUPPORTS OLAP DATA MINING INTERACTIVE DATA EXPLORATION Overview of

More information

Lecture Notes for Chapter 6. Introduction to Data Mining. (modified by Predrag Radivojac, 2017)

Lecture Notes for Chapter 6. Introduction to Data Mining. (modified by Predrag Radivojac, 2017) Lecture Notes for Chapter 6 Introduction to Data Mining by Tan, Steinbach, Kumar (modified by Predrag Radivojac, 27) Association Rule Mining Given a set of transactions, find rules that will predict the

More information

Frequent Itemset Mining

Frequent Itemset Mining ì 1 Frequent Itemset Mining Nadjib LAZAAR LIRMM- UM COCONUT Team (PART I) IMAGINA 17/18 Webpage: http://www.lirmm.fr/~lazaar/teaching.html Email: lazaar@lirmm.fr 2 Data Mining ì Data Mining (DM) or Knowledge

More information

Distributed Mining of Frequent Closed Itemsets: Some Preliminary Results

Distributed Mining of Frequent Closed Itemsets: Some Preliminary Results Distributed Mining of Frequent Closed Itemsets: Some Preliminary Results Claudio Lucchese Ca Foscari University of Venice clucches@dsi.unive.it Raffaele Perego ISTI-CNR of Pisa perego@isti.cnr.it Salvatore

More information

Outline. Fast Algorithms for Mining Association Rules. Applications of Data Mining. Data Mining. Association Rule. Discussion

Outline. Fast Algorithms for Mining Association Rules. Applications of Data Mining. Data Mining. Association Rule. Discussion Outline Fast Algorithms for Mining Association Rules Rakesh Agrawal Ramakrishnan Srikant Introduction Algorithm Apriori Algorithm AprioriTid Comparison of Algorithms Conclusion Presenter: Dan Li Discussion:

More information

Lars Schmidt-Thieme, Information Systems and Machine Learning Lab (ISMLL), University of Hildesheim, Germany

Lars Schmidt-Thieme, Information Systems and Machine Learning Lab (ISMLL), University of Hildesheim, Germany Syllabus Fri. 21.10. (1) 0. Introduction A. Supervised Learning: Linear Models & Fundamentals Fri. 27.10. (2) A.1 Linear Regression Fri. 3.11. (3) A.2 Linear Classification Fri. 10.11. (4) A.3 Regularization

More information

Data Mining. Dr. Raed Ibraheem Hamed. University of Human Development, College of Science and Technology Department of Computer Science

Data Mining. Dr. Raed Ibraheem Hamed. University of Human Development, College of Science and Technology Department of Computer Science Data Mining Dr. Raed Ibraheem Hamed University of Human Development, College of Science and Technology Department of Computer Science 2016 2017 Road map The Apriori algorithm Step 1: Mining all frequent

More information

Association Analysis: Basic Concepts. and Algorithms. Lecture Notes for Chapter 6. Introduction to Data Mining

Association Analysis: Basic Concepts. and Algorithms. Lecture Notes for Chapter 6. Introduction to Data Mining Association Analysis: Basic Concepts and Algorithms Lecture Notes for Chapter 6 Introduction to Data Mining by Tan, Steinbach, Kumar Tan,Steinbach, Kumar Introduction to Data Mining 4/18/2004 1 Association

More information

Reductionist View: A Priori Algorithm and Vector-Space Text Retrieval. Sargur Srihari University at Buffalo The State University of New York

Reductionist View: A Priori Algorithm and Vector-Space Text Retrieval. Sargur Srihari University at Buffalo The State University of New York Reductionist View: A Priori Algorithm and Vector-Space Text Retrieval Sargur Srihari University at Buffalo The State University of New York 1 A Priori Algorithm for Association Rule Learning Association

More information

CSE 5243 INTRO. TO DATA MINING

CSE 5243 INTRO. TO DATA MINING CSE 5243 INTRO. TO DATA MINING Mining Frequent Patterns and Associations: Basic Concepts (Chapter 6) Huan Sun, CSE@The Ohio State University 10/17/2017 Slides adapted from Prof. Jiawei Han @UIUC, Prof.

More information

Associa'on Rule Mining

Associa'on Rule Mining Associa'on Rule Mining Debapriyo Majumdar Data Mining Fall 2014 Indian Statistical Institute Kolkata August 4 and 7, 2014 1 Market Basket Analysis Scenario: customers shopping at a supermarket Transaction

More information

Meelis Kull Autumn Meelis Kull - Autumn MTAT Data Mining - Lecture 05

Meelis Kull Autumn Meelis Kull - Autumn MTAT Data Mining - Lecture 05 Meelis Kull meelis.kull@ut.ee Autumn 2017 1 Sample vs population Example task with red and black cards Statistical terminology Permutation test and hypergeometric test Histogram on a sample vs population

More information

DATA MINING LECTURE 4. Frequent Itemsets and Association Rules

DATA MINING LECTURE 4. Frequent Itemsets and Association Rules DATA MINING LECTURE 4 Frequent Itemsets and Association Rules This is how it all started Rakesh Agrawal, Tomasz Imielinski, Arun N. Swami: Mining Association Rules between Sets of Items in Large Databases.

More information

Introduction to Data Mining

Introduction to Data Mining Introduction to Data Mining Lecture #12: Frequent Itemsets Seoul National University 1 In This Lecture Motivation of association rule mining Important concepts of association rules Naïve approaches for

More information

The Market-Basket Model. Association Rules. Example. Support. Applications --- (1) Applications --- (2)

The Market-Basket Model. Association Rules. Example. Support. Applications --- (1) Applications --- (2) The Market-Basket Model Association Rules Market Baskets Frequent sets A-priori Algorithm A large set of items, e.g., things sold in a supermarket. A large set of baskets, each of which is a small set

More information

Association Rule Mining on Web

Association Rule Mining on Web Association Rule Mining on Web What Is Association Rule Mining? Association rule mining: Finding interesting relationships among items (or objects, events) in a given data set. Example: Basket data analysis

More information

COMP 5331: Knowledge Discovery and Data Mining

COMP 5331: Knowledge Discovery and Data Mining COMP 5331: Knowledge Discovery and Data Mining Acknowledgement: Slides modified by Dr. Lei Chen based on the slides provided by Tan, Steinbach, Kumar And Jiawei Han, Micheline Kamber, and Jian Pei 1 10

More information

Frequent Itemset Mining

Frequent Itemset Mining 1 Frequent Itemset Mining Nadjib LAZAAR LIRMM- UM IMAGINA 15/16 2 Frequent Itemset Mining: Motivations Frequent Itemset Mining is a method for market basket analysis. It aims at finding regulariges in

More information

Un nouvel algorithme de génération des itemsets fermés fréquents

Un nouvel algorithme de génération des itemsets fermés fréquents Un nouvel algorithme de génération des itemsets fermés fréquents Huaiguo Fu CRIL-CNRS FRE2499, Université d Artois - IUT de Lens Rue de l université SP 16, 62307 Lens cedex. France. E-mail: fu@cril.univ-artois.fr

More information

Introduction to Data Mining, 2 nd Edition by Tan, Steinbach, Karpatne, Kumar

Introduction to Data Mining, 2 nd Edition by Tan, Steinbach, Karpatne, Kumar Data Mining Chapter 5 Association Analysis: Basic Concepts Introduction to Data Mining, 2 nd Edition by Tan, Steinbach, Karpatne, Kumar 2/3/28 Introduction to Data Mining Association Rule Mining Given

More information

FREQUENT PATTERN SPACE MAINTENANCE: THEORIES & ALGORITHMS

FREQUENT PATTERN SPACE MAINTENANCE: THEORIES & ALGORITHMS FREQUENT PATTERN SPACE MAINTENANCE: THEORIES & ALGORITHMS FENG MENGLING SCHOOL OF ELECTRICAL & ELECTRONIC ENGINEERING NANYANG TECHNOLOGICAL UNIVERSITY 2009 FREQUENT PATTERN SPACE MAINTENANCE: THEORIES

More information

Approximate counting: count-min data structure. Problem definition

Approximate counting: count-min data structure. Problem definition Approximate counting: count-min data structure G. Cormode and S. Muthukrishhan: An improved data stream summary: the count-min sketch and its applications. Journal of Algorithms 55 (2005) 58-75. Problem

More information

Descrip9ve data analysis. Example. Example. Example. Example. Data Mining MTAT (6EAP)

Descrip9ve data analysis. Example. Example. Example. Example. Data Mining MTAT (6EAP) 3.9.2 Descrip9ve data analysis Data Mining MTAT.3.83 (6EAP) hp://courses.cs.ut.ee/2/dm/ Frequent itemsets and associa@on rules Jaak Vilo 2 Fall Aims to summarise the main qualita9ve traits of data. Used

More information

Effective Elimination of Redundant Association Rules

Effective Elimination of Redundant Association Rules Effective Elimination of Redundant Association Rules James Cheng Yiping Ke Wilfred Ng Department of Computer Science and Engineering The Hong Kong University of Science and Technology Clear Water Bay,

More information

Maintaining Frequent Itemsets over High-Speed Data Streams

Maintaining Frequent Itemsets over High-Speed Data Streams Maintaining Frequent Itemsets over High-Speed Data Streams James Cheng, Yiping Ke, and Wilfred Ng Department of Computer Science Hong Kong University of Science and Technology Clear Water Bay, Kowloon,

More information

Causal Discovery Methods Using Causal Probabilistic Networks

Causal Discovery Methods Using Causal Probabilistic Networks ausal iscovery Methods Using ausal Probabilistic Networks MINFO 2004, T02: Machine Learning Methods for ecision Support and iscovery onstantin F. liferis & Ioannis Tsamardinos iscovery Systems Laboratory

More information

Efficient search of association rules with Fisher s exact test

Efficient search of association rules with Fisher s exact test fficient search of association rules with Fisher s exact test W. Hämäläinen Abstract limination of spurious and redundant rules is an important problem in the association rule discovery. Statistical measure

More information

Mining Molecular Fragments: Finding Relevant Substructures of Molecules

Mining Molecular Fragments: Finding Relevant Substructures of Molecules Mining Molecular Fragments: Finding Relevant Substructures of Molecules Christian Borgelt, Michael R. Berthold Proc. IEEE International Conference on Data Mining, 2002. ICDM 2002. Lecturers: Carlo Cagli

More information

Frequent Itemset Mining

Frequent Itemset Mining ì 1 Frequent Itemset Mining Nadjib LAZAAR LIRMM- UM COCONUT Team IMAGINA 16/17 Webpage: h;p://www.lirmm.fr/~lazaar/teaching.html Email: lazaar@lirmm.fr 2 Data Mining ì Data Mining (DM) or Knowledge Discovery

More information

On Compressing Frequent Patterns

On Compressing Frequent Patterns On Compressing Frequent Patterns Dong Xin Jiawei Han Xifeng Yan Hong Cheng Department of Computer Science, University of Illinois at Urbana-Champaign, Urbana, IL 680, USA Abstract A major challenge in

More information

Interesting Patterns. Jilles Vreeken. 15 May 2015

Interesting Patterns. Jilles Vreeken. 15 May 2015 Interesting Patterns Jilles Vreeken 15 May 2015 Questions of the Day What is interestingness? what is a pattern? and how can we mine interesting patterns? What is a pattern? Data Pattern y = x - 1 What

More information

On Condensed Representations of Constrained Frequent Patterns

On Condensed Representations of Constrained Frequent Patterns Under consideration for publication in Knowledge and Information Systems On Condensed Representations of Constrained Frequent Patterns Francesco Bonchi 1 and Claudio Lucchese 2 1 KDD Laboratory, ISTI Area

More information

2.6 Complexity Theory for Map-Reduce. Star Joins 2.6. COMPLEXITY THEORY FOR MAP-REDUCE 51

2.6 Complexity Theory for Map-Reduce. Star Joins 2.6. COMPLEXITY THEORY FOR MAP-REDUCE 51 2.6. COMPLEXITY THEORY FOR MAP-REDUCE 51 Star Joins A common structure for data mining of commercial data is the star join. For example, a chain store like Walmart keeps a fact table whose tuples each

More information

Basic Data Structures and Algorithms for Data Profiling Felix Naumann

Basic Data Structures and Algorithms for Data Profiling Felix Naumann Basic Data Structures and Algorithms for 8.5.2017 Overview 1. The lattice 2. Apriori lattice traversal 3. Position List Indices 4. Bloom filters Slides with Thorsten Papenbrock 2 Definitions Lattice Partially

More information

.. Cal Poly CSC 466: Knowledge Discovery from Data Alexander Dekhtyar..

.. Cal Poly CSC 466: Knowledge Discovery from Data Alexander Dekhtyar.. .. Cal Poly CSC 4: Knowledge Discovery from Data Alexander Dekhtyar.. Data Mining: Mining Association Rules Examples Course Enrollments Itemset. I = { CSC3, CSC3, CSC40, CSC40, CSC4, CSC44, CSC4, CSC44,

More information

Pattern Space Maintenance for Data Updates. and Interactive Mining

Pattern Space Maintenance for Data Updates. and Interactive Mining Pattern Space Maintenance for Data Updates and Interactive Mining Mengling Feng, 1,3,4 Guozhu Dong, 2 Jinyan Li, 1 Yap-Peng Tan, 1 Limsoon Wong 3 1 Nanyang Technological University, 2 Wright State University

More information

EFFICIENT MINING OF WEIGHTED QUANTITATIVE ASSOCIATION RULES AND CHARACTERIZATION OF FREQUENT ITEMSETS

EFFICIENT MINING OF WEIGHTED QUANTITATIVE ASSOCIATION RULES AND CHARACTERIZATION OF FREQUENT ITEMSETS EFFICIENT MINING OF WEIGHTED QUANTITATIVE ASSOCIATION RULES AND CHARACTERIZATION OF FREQUENT ITEMSETS Arumugam G Senior Professor and Head, Department of Computer Science Madurai Kamaraj University Madurai,

More information

Algorithmic Methods of Data Mining, Fall 2005, Course overview 1. Course overview

Algorithmic Methods of Data Mining, Fall 2005, Course overview 1. Course overview Algorithmic Methods of Data Mining, Fall 2005, Course overview 1 Course overview lgorithmic Methods of Data Mining, Fall 2005, Course overview 1 T-61.5060 Algorithmic methods of data mining (3 cp) P T-61.5060

More information

Pushing Tougher Constraints in Frequent Pattern Mining

Pushing Tougher Constraints in Frequent Pattern Mining Pushing Tougher Constraints in Frequent Pattern Mining Francesco Bonchi 1 and Claudio Lucchese 2 1 Pisa KDD Laboratory, ISTI - C.N.R., Area della Ricerca di Pisa, Italy 2 Department of Computer Science,

More information

Exercise 1. min_sup = 0.3. Items support Type a 0.5 C b 0.7 C c 0.5 C d 0.9 C e 0.6 F

Exercise 1. min_sup = 0.3. Items support Type a 0.5 C b 0.7 C c 0.5 C d 0.9 C e 0.6 F Exercise 1 min_sup = 0.3 Items support Type a 0.5 C b 0.7 C c 0.5 C d 0.9 C e 0.6 F Items support Type ab 0.3 M ac 0.2 I ad 0.4 F ae 0.4 F bc 0.3 M bd 0.6 C be 0.4 F cd 0.4 C ce 0.2 I de 0.6 C Items support

More information

CSE 5243 INTRO. TO DATA MINING

CSE 5243 INTRO. TO DATA MINING CSE 5243 INTRO. TO DATA MINING Advanced Frequent Pattern Mining & Locality Sensitivity Hashing Huan Sun, CSE@The Ohio State University /7/27 Slides adapted from Prof. Jiawei Han @UIUC, Prof. Srinivasan

More information

CS 412 Intro. to Data Mining

CS 412 Intro. to Data Mining CS 412 Intro. to Data Mining Chapter 6. Mining Frequent Patterns, Association and Correlations: Basic Concepts and Methods Jiawei Han, Computer Science, Univ. Illinois at Urbana -Champaign, 2017 1 2 3

More information

Machine Learning: Pattern Mining

Machine Learning: Pattern Mining Machine Learning: Pattern Mining Information Systems and Machine Learning Lab (ISMLL) University of Hildesheim Wintersemester 2007 / 2008 Pattern Mining Overview Itemsets Task Naive Algorithm Apriori Algorithm

More information

Discovering Threshold-based Frequent Closed Itemsets over Probabilistic Data

Discovering Threshold-based Frequent Closed Itemsets over Probabilistic Data Discovering Threshold-based Frequent Closed Itemsets over Probabilistic Data Yongxin Tong #1, Lei Chen #, Bolin Ding *3 # Department of Computer Science and Engineering, Hong Kong Univeristy of Science

More information

CS5112: Algorithms and Data Structures for Applications

CS5112: Algorithms and Data Structures for Applications CS5112: Algorithms and Data Structures for Applications Lecture 19: Association rules Ramin Zabih Some content from: Wikipedia/Google image search; Harrington; J. Leskovec, A. Rajaraman, J. Ullman: Mining

More information

arxiv: v1 [cs.db] 31 Dec 2011

arxiv: v1 [cs.db] 31 Dec 2011 Mining Flipping Correlations from Large Datasets with Taxonomies MarinaBarsky SangkyumKim TimWeninger JiaweiHan Univ. ofvictoria,bc,canada,marina barsky@gmail.com Univ.ofIllinoisatUrbana-Champaign, {kim71,weninger1,hanj}@illinois.edu

More information

Association Analysis. Part 1

Association Analysis. Part 1 Association Analysis Part 1 1 Market-basket analysis DATA: A large set of items: e.g., products sold in a supermarket A large set of baskets: e.g., each basket represents what a customer bought in one

More information

Handling a Concept Hierarchy

Handling a Concept Hierarchy Food Electronics Handling a Concept Hierarchy Bread Milk Computers Home Wheat White Skim 2% Desktop Laptop Accessory TV DVD Foremost Kemps Printer Scanner Data Mining: Association Rules 5 Why should we

More information

Density-Based Clustering

Density-Based Clustering Density-Based Clustering idea: Clusters are dense regions in feature space F. density: objects volume ε here: volume: ε-neighborhood for object o w.r.t. distance measure dist(x,y) dense region: ε-neighborhood

More information

Frequent Itemset Mining

Frequent Itemset Mining Frequent Itemset Mining prof. dr Arno Siebes Algorithmic Data Analysis Group Department of Information and Computing Sciences Universiteit Utrecht Battling Size The previous time we saw that Big Data has

More information

Knowledge Discovery and Data Mining I

Knowledge Discovery and Data Mining I Ludwig-Maximilians-Universität München Lehrstuhl für Datenbanksysteme und Data Mining Prof. Dr. Thomas Seidl Knowledge Discovery and Data Mining I Winter Semester 2018/19 Agenda 1. Introduction 2. Basics

More information

PFPM: Discovering Periodic Frequent Patterns with Novel Periodicity Measures

PFPM: Discovering Periodic Frequent Patterns with Novel Periodicity Measures PFPM: Discovering Periodic Frequent Patterns with Novel Periodicity Measures Philippe Fournier-Viger a,, Chun-Wei Lin b, Quang-Huy Duong d, Thu-Lan Dam d, Lukáš Ševčík e, Dominik Uhrin e, Miroslav Voznak

More information

1 Frequent Pattern Mining

1 Frequent Pattern Mining Decision Support Systems MEIC - Alameda 2010/2011 Homework #5 Due date: 31.Oct.2011 1 Frequent Pattern Mining 1. The Apriori algorithm uses prior knowledge about subset support properties. In particular,

More information

10/19/2017 MIST.6060 Business Intelligence and Data Mining 1. Association Rules

10/19/2017 MIST.6060 Business Intelligence and Data Mining 1. Association Rules 10/19/2017 MIST6060 Business Intelligence and Data Mining 1 Examples of Association Rules Association Rules Sixty percent of customers who buy sheets and pillowcases order a comforter next, followed by

More information

Mining Free Itemsets under Constraints

Mining Free Itemsets under Constraints Mining Free Itemsets under Constraints Jean-François Boulicaut Baptiste Jeudy Institut National des Sciences Appliquées de Lyon Laboratoire d Ingénierie des Systèmes d Information Bâtiment 501 F-69621

More information

CSE-4412(M) Midterm. There are five major questions, each worth 10 points, for a total of 50 points. Points for each sub-question are as indicated.

CSE-4412(M) Midterm. There are five major questions, each worth 10 points, for a total of 50 points. Points for each sub-question are as indicated. 22 February 2007 CSE-4412(M) Midterm p. 1 of 12 CSE-4412(M) Midterm Sur / Last Name: Given / First Name: Student ID: Instructor: Parke Godfrey Exam Duration: 75 minutes Term: Winter 2007 Answer the following

More information

Descriptive data analysis. E.g. set of items. Example: Items in baskets. Sort or renumber in any way. Example

Descriptive data analysis. E.g. set of items. Example: Items in baskets. Sort or renumber in any way. Example .3.6 Descriptive data analysis Data Mining MTAT.3.83 (6EAP) Frequent itemsets and association rules Jaak Vilo 26 Spring Aims to summarise the main qualitative traits of data. Used mainly for discovering

More information

Mining Statistically Important Equivalence Classes and Delta-Discriminative Emerging Patterns

Mining Statistically Important Equivalence Classes and Delta-Discriminative Emerging Patterns Mining Statistically Important Equivalence Classes and Delta-Discriminative Emerging Patterns Jinyan Li Institute for Infocomm Research (I 2 R) & Nanyang Technological University, Singapore jyli@ntu.edu.sg

More information

OPPA European Social Fund Prague & EU: We invest in your future.

OPPA European Social Fund Prague & EU: We invest in your future. OPPA European Social Fund Prague & EU: We invest in your future. Frequent itemsets, association rules Jiří Kléma Department of Cybernetics, Czech Technical University in Prague http://ida.felk.cvut.cz

More information

Assignment 7 (Sol.) Introduction to Data Analytics Prof. Nandan Sudarsanam & Prof. B. Ravindran

Assignment 7 (Sol.) Introduction to Data Analytics Prof. Nandan Sudarsanam & Prof. B. Ravindran Assignment 7 (Sol.) Introduction to Data Analytics Prof. Nandan Sudarsanam & Prof. B. Ravindran 1. Let X, Y be two itemsets, and let denote the support of itemset X. Then the confidence of the rule X Y,

More information

Mining Approximative Descriptions of Sets Using Rough Sets

Mining Approximative Descriptions of Sets Using Rough Sets Mining Approximative Descriptions of Sets Using Rough Sets Dan A. Simovici University of Massachusetts Boston, Dept. of Computer Science, 100 Morrissey Blvd. Boston, Massachusetts, 02125 USA dsim@cs.umb.edu

More information

Computational Frameworks. MapReduce

Computational Frameworks. MapReduce Computational Frameworks MapReduce 1 Computational challenges in data mining Computation-intensive algorithms: e.g., optimization algorithms, graph algorithms Large inputs: e.g., web, social networks data.

More information

Frequent Subtree Mining on the Automata Processor: Challenges and Opportunities

Frequent Subtree Mining on the Automata Processor: Challenges and Opportunities Frequent Subtree Mining on the Automata Processor: hallenges and Opportunities Elaheh Sadredini, Reza Rahimi, Ke Wang, Kevin Skadron Department of omputer Science University of Virginia Presenting in:

More information

CS246 Final Exam. March 16, :30AM - 11:30AM

CS246 Final Exam. March 16, :30AM - 11:30AM CS246 Final Exam March 16, 2016 8:30AM - 11:30AM Name : SUID : I acknowledge and accept the Stanford Honor Code. I have neither given nor received unpermitted help on this examination. (signed) Directions

More information

Data mining, 4 cu Lecture 5:

Data mining, 4 cu Lecture 5: 582364 Data mining, 4 cu Lecture 5: Evaluation of Association Patterns Spring 2010 Lecturer: Juho Rousu Teaching assistant: Taru Itäpelto Evaluation of Association Patterns Association rule algorithms

More information

Mining Rank Data. Sascha Henzgen and Eyke Hüllermeier. Department of Computer Science University of Paderborn, Germany

Mining Rank Data. Sascha Henzgen and Eyke Hüllermeier. Department of Computer Science University of Paderborn, Germany Mining Rank Data Sascha Henzgen and Eyke Hüllermeier Department of Computer Science University of Paderborn, Germany {sascha.henzgen,eyke}@upb.de Abstract. This paper addresses the problem of mining rank

More information

CS738 Class Notes. Steve Revilak

CS738 Class Notes. Steve Revilak CS738 Class Notes Steve Revilak January 2008 May 2008 Copyright c 2008 Steve Revilak. Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation

More information