Lieu theo tir OLAP va cac nguen MI lieu quan he.

Một phần của tài liệu Khai phá dữ liệu trong SQL server 2012.PDF (Trang 37)

- Cay quyet djnh du down met khach hang cu the se mua met son pham hay kh6ng

lieu theo tir OLAP va cac nguen MI lieu quan he.

Trang 36190 A11278 — Doan Thinh Cong

KHAI PHA DU LIEU TRONG SQL SERVER 2012

Vi du: sir dung thuat toan nay de du doan bin hang va lqi nhui'n dui vao cac dit

lieu qua khir trong 1 cube

Sir dung thuat toan Time Series do the chqn mot hoac nhieu bien de du doin (nhung cac bien phai la lien tuc). CO the c6 nhieu trueng hqp cho mai mo hinh. Tap

cac tnrtmg hqp xac dinh vi tri cua met thorn, nhu la ngiy thing khi xem viec bin hang

thong qua vai thing hoc vai nam truerc.

Mot twang hqp co the bao gem met tap cac bien (vi du nhu ban hang tai cac cira hang Ichic nhau). Thuat town nay co the sir dung sir tuong quan cua thay doi bien so

trong du doin ctia no.

Vi du.: bin hang trtrem kia tai met cira hang co the rat heru ich trong viec du bao

bin hang hien tai tai nhimg cira hang.

3.2.6. Microsoft Association Rules

Th4't toan nay duqc thiet ke de sir dung phan tich gio hang thi twang (basket market) ten str giao dich cita khach hang. Nher thult town phan tich lust ket hqp co the biet dtrqc nhcmg sin phim nao thuerng dtrqc bin ding yeti nhau va lam the nao met san pham dic Wet duqc ban cling veri nhemg san phim khic. Vi du 5% so ichich mua laptop, chuet khong day ding v6i de tan nhiet va 90% cim ithimg khich hang nay da mua laptop, chuot khong day thi cling se mua de tan nhiet.

Thuat toan phan tich lust ket hqp, thuat toan nay se xet mai cap thuec tinh/gia tri la mot item. Met Itemset la met tap hqp cac item trong 1 giao dich (transaction) don le.Thuat toan se quat qua cac tap dit lieu, tim kiem cac tap Itemset xuat hien trong nhieu giao dich. Tham chieu Support se dinh nghia bao nhieu transaction ma itemset

se xui't hien tank khi no duqc go:pi la quan tong.

Vi du ve met itemset phi) bien:

{Gender="Male", Marital Status="Married", Age="30-35"}

Trong Itemset nay co cac Item: Gender=male; MaritalStatus =Married; Age=30- 35

Trong Item Gender la thuec tinh, male dirge goi la gia trf cita thuec tinh gender Thuat toan nay cling tim ra 141 ket hqp giara cac Iternset.Vi du, Mot luit kat hqp 05 ding X,Y=>Z. Khi ca X,Y,Z deu la Itemset phe bien thi to not ring Z duqc du (loan to X,Y.

Trang 371 90 A11278 — Doan Thanh Cling

KHAI pHA DC! LIEU TRONG SQL SERVER 2012

Mqt dac tinh quan tong nita cita phan tich tat ket hop d6 la Probability (xac sat). Xic suat cua quy tee ket hop A=>B duqc tinh town bang each sir dung Support cita itemset (A,B) chia cho Support cita itemset A. Xic sat nay duqc goi la dq tin cay trong nhitng nghien ciru ctia khai phi du

3.2.7. Microsoft Neural Network

That town Neural Network tao cac m8 hinh khai thic fel h8i quy vi phin loaf bang each xay dung da lop perceptim cfra cac flown. Citing nhu that toan cay quyet djnh, dua ra mei tinh twig cita thuOc tinh co the do doin. That tan Neural Network tinh toan kha nang co the cita mei trang thai ce, the Gila thuOc tinh dau vao. That toan Neural Network se xir 1$' tan the cac trtrerng hop. So lap di lap tai so sinh the du (loan phan loaf cfra cac truerng hop vfri so phan loci ca cac truerng dA biet. Sai se tir so phin loci ban diu (dm phep lap ban dAu) cita toan bo cac twang hqp duqc tra ve network va dtrqc sir dung de thay aM so thoc thi cita network cho cac phop lap ke theo,v.v... co the sau do sir dung nheng kha nang nay de do doin ket qua cita cac thuOc tinh do doan, dua tren thuete tinh vac).

Mqt so khach biet chink gifra that town Neural Network va that town cay guy& djnh la cac lcien thirc xir li la nhi'mg tham so network ton uu nhim lam nhe !that cac loi co the trong khi cay quyet djnh tach cac !at, moc dich de eve dal h6a th8ng tin co 10. That town nay he trq ca thuqc tinh raj rac va IS tic.

3.2.8. Microsoft Linear Regression

Microsoft Linear Regression la met cau hinh co the cita that town Microsoft Decision Trees, thu duqc bang each vo hieu hew chia tach (cac tong thirc hqi quy town bq duqc xay (long trong met nut gec duy nhat). That town nay he trq cac do down cho cac thuOc firth lien tic.

3.2.9. Microsoft Logistic Regression

Microsoft Logistic Regression la mgt cau hinh co the cita that town Microsoft Neural Network, thu duqc bang each loci bo cac lop An. That town nay he trq cac do doom cita ca hai thuqc tinh red rac va lien tic.

3.3. Nguyen tic chon thu$t toin

SQL Server bao gam nhfrng that tan sau: (adsbygoogle = window.adsbygoogle || []).push({});

Trang 381 90 A11278 — Doan Thinh Ding

KHAI PHA DC! LIEU TRONG SQL SERVER 2012

Thuat town phin loci: du doin mot hok nhieu bien rai roc (khong lion tic), dua ten cac thu6c firth trong Lap de lieu (Microsoft Decision Trees Algorithm).

Thu* toan hoi quy: du doin moat hoc nhieu hien lien tic, kieu nhu nhiing lqi nhuan va nhffg ban that, dua ten cac thuetc tinh khic nhau cith tap hqp dit lieu (Microsoft Time Series Algorithm).

Thuat tan phin down: chia dii lieu thinh hai nhom, hok cac lien cung, hok cac danh mac c6 thuOc tinh gicang nhau (Microsoft Clustering Algorithm). Thuat Wan ket hqp: tim nhiing su tuang quan gicra cac thuOc tinh khach nhau trong met tap hqp du lieu. ling dung pile hien nhat dua loci thuat toan nay 11 tao ra cac lust ket hqp, do the dirk dimg trong market basket (Microft Association Algorithm).

Thuat Wan phan tich tien trinh: hang kit nhiing tien trinh thirerng xay ra hoc it xay ra trong der lieu (Microsoft Sequence Clustering Algorithm).

Chan moat thuat toan dimg de sir dung cho cac nghiep va Hong biet la mot nhiem via kh6 lchAn. Khi to c6 the sir dung cac thuat toan khic nhau de thuc thi ding met nghiep vu, moi thuat Wan tao ra met ket qua khich nhau, va mot vai thuat Wan c6 the tao ra nhieu han met ket qua.

Vi du 1 c6 the sir dung thuat than Microsoft Decision Trees kh8ng chi de di;

down ma din la met ckh de giam s6 litmg cot trong dataset, bai vi cay quyet djnh co the xk dinh cac got ma khong anh hitmg den me hinh khai thic cuei ding.

Ta cling kung phai sir dung vac thuat than doc lap trong giii phip khai thic dii lieu don gian, c6 the sir dung vai thuat toan de khao sat du lieu, va sau d6 sir dung cac thuat tan de du down kat qua reri roc dta ten du lieu nay.

Vi du 2: c6 the sir dung thuat man gom nhom, nhan ra cac mau, dua dii lieu vao

nhom dOng that, va sau d6 sir dung cac ket qua de tao ra mo hinh cay quyet Binh tot

han.

Vi du 3: bang each sir dung thuat toan cay hal quy de ley thong tin du doin ve tai

chinh, va thuat Wan dua teen luat de dux thi viec khao sat thi throng.

Cac mo hinh khai thic co the du doom cac gin tri, dua ra bang tom tit dii lieu, va tim ra su Prong quan An. De giup cho lta chop thuat toan cho giai phip khai thic chi

Trang 39190 A11278 — Doan Thanh Ding

KHAI PHA Dir LIEU TRONG SQL SERVER 2012

lieu. Bang duoi day cung cap cac gqi y cho vi'ec Ira chcon thuat town nio cho cac clang viec cu the nio:

out; HI@ tu.in str dung

Du Joan thuqc tudi hen me Thuat twin Microsoft Decision Trees VI

VI du: du doin doanh thu nam tiep theo Thu$t toin Microsoft Time Series

Tim nhom lulling di twig trong cac Thu$t town Microsoft Association giao thirc thuc hien Thu* toan Microsoft Decision Trees Vi du: sir dung phan tick thj throng de

dua them cac san phim cho khich hang

Trang 40190 A11278 — Doan Thanh Cong

UNG DUNG KHAI PHA Doi MU SQL SERVER 2012

CHU'ONG 4. UNG DUNG KHAI PHA DIY LI€U SQL SERVER 2012 4.1. Gioi thieu ve Business Intelligence Development Studio 4.1. Gioi thieu ve Business Intelligence Development Studio

Một phần của tài liệu Khai phá dữ liệu trong SQL server 2012.PDF (Trang 37)