270 likes | 390 Views
Mi ning Frequent Episodes for relating Financial Events and Stock Trends. Anny Ng and Ada Wai-chee Fu PAKDD 2003 報告者: Ming Jing Tsai. Definition. Events : financial news ,political … e 1 ,e 2 ,e 3 … .,e k : event types day record D i :{e i1 ,e i2 ,e i3 … .,e ik }
E N D
Mining Frequent Episodes for relating Financial Events and Stock Trends Anny Ng and Ada Wai-chee Fu PAKDD 2003 報告者:Ming Jing Tsai date:2004/03/05
Definition • Events : financial news ,political… • e1,e2,e3….,ek : event types • day record Di:{ei1,ei2,ei3….,eik} • Episode:{e1,e2,e3….,ek},has at least two elements and at least one ej is a stock event type • Window = x days
Definition • Window frequency:number of windows that contains an event type • DB frequency:number of occurrences of an event type in DB • Frequency of an episode (ex) • number of windows • the first day of window contains at least one of the event types in episode.
Header in descending db frequencies order Event_set pair <(firstday) ,(remaining day)> sorted in the descending db frequencies node<E:C:B>: E :event type ,c :counts ,b :binary bit Construct event tree
Pruning method • window frequencies < min_sup • Remove duplicate event type in both firstday part and remaining day part
An Event database Window = 3,min_sup =3 Db frequencies<a:2,b:3,c:2,d:2>
windows Window = 3,min_sup =3 Ordered frequent event type<b,a,c,d> Window frequencies<a:5,b:5,c:5,d:6>
{null} {b:1:0} {a:1:0} {a:1:1} {c:1:0} {c:1:1} {b:1:1} {d:1:1}
{null} {b:2:0} {a:1:0} {d:1:0} {a:1:1} {c:1:0} {b:1:1} {d:1:1} {c:1:1} {b:1:1} {a:1:1} {d:1:1} {c:1:1}
{null} {b:3:0} {a:1:0} {d:1:0} {a:1:1} {c:1:0} {b:1:1} {d:1:1} {c:1:1} {b:1:1} {a:1:1} {d:1:1} {c:1:1}
{null} {b:3:0} {a:1:0} {d:1:0} {a:2:1} {c:1:0} {b:1:1} {d:1:1} {c:2:1} {b:1:1} {a:1:1} {d:1:1} {d:1:1} {c:1:1}
{null} {b:3:0} {a:2:0} {d:1:0} {a:2:1} {c:2:0} {b:1:1} {d:1:1} {c:2:1} {b:1:1} {a:1:1} {d:1:1} {d:1:1} {d:1:1} {c:1:1}
{null} {b:3:0} {a:2:0} {d:2:0} {a:2:1} {c:2:0} {b:1:1} {d:1:1} {c:2:1} {b:1:1} {a:1:1} {d:1:1} {d:1:1} {d:1:1} {c:1:1}
Mining frequent episode • Header table{h0,h1,…..,hH} • Mining recursively each of the linked list kept at the header table • from bottom to top • Conditional path can build conditional event tree • Object 1:found frequent episodes of form {a} ∪{hi} • first-part frequencies • Object 2:found frequent episodes that contain hi and at least two other event types • Db frequencies
Traverse conditional path • Remove invalid event types • Adjust counts of nodes above hi in the path to be equal to that of hi • If hi is in the firstdays part, then move all event types in the remainingdays part to the firstdays part • Remove hi from the path
Generate frequent episode • When a conditional event tree contains only a single path • Any subset of firstpart ∪ event base set • Any Subsets of firstpart ∪ Any Subsets of remainingpart ∪ event base set
Mining Header d min_sup =3 event base set {d} • <(a:1,c:1),(b:1)> • <(b:1),()> • <(b:1,a:1,c:1),()> • <(b:1),(a:1,c:1)> • <(a:1,c:1),()> db frequency:{<b:4,a:4,c:4>} First_part frequency:{<b:3,a:3,c:3>} Frequent episode :{bd,ad,cd}
Recursively Mining Header c event base set {cd} • <(a:1,b:1),()> • <(b:1,a:1),()> • <(b:1),(a:1)> • <(a:1),()> <(a:1,c:1),(b:1)> <(b:1),()> <(b:1,a:1,c:1),()> <(b:1),(a:1,c:1)> <(a:1,c:1),()> db frequency:{<b:3,a:4>} First_part frequency:{<b:3,a:3>} Frequent episode :{bcd ,acd}
Recursively Mining Header a event base set {acd} • <(b:1),()> • <(b:1),()> • <(b:1),()> <(a:1,b:1),()> <(b:1,a:1),()> <(b:1),(a:1)> <(a:1),()> db frequency:{<b:3>} First_part frequency:{<b:3>} Frequent episode :{bacd}
Mining Header c min_sup =3 event base set {c} • <(b:1),(a:1)> • <(a:1,b:1),()> • <(b:1),(a:1)> • <(a:1),()> db frequency:{<b:3,a:4>} First_part frequency:{<b:3,a:2>} Frequent episode :{bc}
Recursively Mining Header a min_sup =3 event base set {ac} • <(b:1),()> • <(b:1),()> • <(b:1),()> db frequency:{<b:3>} First_part frequency:{<b:3>} Frequent episode :{bac}
Mining Header a min_sup =3 event base set {a} • <(b:1),()> • <(b:1),()> • <(b:1),()> db frequency:{<b:3>} First_part frequency:{<b:3>} Frequent episode :{ba}
Experiment (real data) • News event from a internet • 121 event types • 757 days • Stock data • Dow Jones ,Nasdaq ,Hang Seng , 12 top local companies