Urongwa muDhina Mining

Kuenzanisa inzira yekumigodhi inoshandiswa iyo inopa zvikamu kuunganidzwa kwedata kuitira kuti ubatsire mune zvakafanotaurwa zvakajeka uye kuongorora. Iyo inonziwo dzimwe nguva inonzi Decision Tree , kusarudzwa ndeimwe yemhando dzakawanda dzakarongwa kuita kuti kuongororwa kwemashoko makuru e datasets agone.

Sei Tichifanira Kuiswa?

Databases yakawanda kwazvo iri kuve yakajairika munyika yanhasi ye "guru data." Fungidzira dhesikisi yakawanda yeetababytes yedata -a terabyte imwe trillion bytes yedata.

Facebook chete inokonzera 600 terabytes yedheta idzva zuva rimwe nerimwe (segore ra2014, nguva yekupedzisira iyo yakashambadza tsanangudzo idzi). Icho chinetso chikuru chemashoko makuru ndechekuti inganzwisisa sei.

Uye iyo yakawanda haisiyo dambudziko chete: nhoroondo huru inowanzove yakasiyana, isinganzwisisiki uye inokurumidza-kuchinja. Funga nezvemavhidhiyo nemavhidhiyo data, zvinyorwa zvezvemagariro evanhu, 3D data kana geospatial data. Nyaya iyi yedata haisi nyore kuiswa kana kurongeka.

Kutarisana nechinetso ichi, mararamiro ezvimwe zviito zvekutsvaga ruzivo rwakakosha yakagadzirwa, pakati pavo maitiro .

Nzira Yokuenzanisa Inoshanda Sei?

Pangozi yekufambisa kure mu tech-talk, regai tikurukure kuti kushambadza kunoshanda sei. Chinangwa ndechokuita sarudzo yemitemo inopindura mubvunzo, kuita sarudzo, kana kufanotaura maitiro.Kutanga, sarudzo yekudzidzira dambudziko yakagadzirwa ine chitubu chezvimiridzo uye pamwe nemugumisiro.

Basa rekugadzirisa zvigadziriswa nderokuona kuti iyo yakagadziriswa inosvika pakuguma kwayo.

Muenzaniso : Zvichida kambani yekadhi rechikwereti iri kuedza kutsvaga kuti ndeipi tarisiro inofanirwa kugamuchira kadhi rechikwereti.

Izvi zvingangodaro yakagadzirwa yekudzidzira deta:

Kudzidzisa Data
Zita Age Gender Mari Yegore negore Credit Card Offer
John Doe 25 M $ 39,500 Aihwa
Jane Doe 56 F $ 125,000 Ehe

Izwi rokuti "predictor" mazera Age , Gender , uye Annual Income inoratidza kukosha kwe "predictor" " Credit Card Offer . Mukurudziro yakagadzirirwa, chirevo chekugadzirira chinozivikanwa. Icho chinonzi algorithm zvino chinoedza kuziva kuti kukosha kwehupfungwa hwekufananidzira kwakasvika sei: ndehupi hukama huripo pakati pekufungidzira uye chisarudzo? Ichakudziridza chirevo chekufanotaura kwemitemo, kazhinji kacho chirevo che IF / THEN, somuenzaniso:

KANA (Zera> 18 OR Age <75) NOKUDZIDZWA KWAKAVIMA> 40 000 ZVINO KUDZIDZA KUNA KUDHA = hongu

Zviri pachena, uyu muenzaniso wakajeka, uye sargorithm inoda kudarika yakakura kudarika sampling kupfuura zvinyorwa zviviri zvinoratidzwa pano. Uyezve, mitemo yekufanotaurira inogona kunge yakanyanya kuoma, kusanganisira mitemo-inotora zvinyorwa.

Zvadaro, iyo algorithm inopiwa "kufanirwa kusarudzwa" kwe data kuti iongorore, asi izvi zvinenge zvisina chirevo chekufananidzira (kana chisarudzo):

Predictor Data
Zita Age Gender Mari Yegore negore Credit Card Offer
Jack Frost 42 M $ 88,000
Mary Murray 16 F $ 0

Izvi zvinotarisa dambudziko rinobatsira kufungidzira kuve kwechokwadi kwemitemo yekufanotaura, uye mitemo yacho inobva yatora kusvika kusvikira musikoni anofunga kuti huporofita hunobudirira uye hunobatsira.

Zuva Nezuva Mienzaniso yekutsvaga

Kuenzanisa, nedzimwe nzira dzekugadzira migodhi, zvinokonzera ruzivo rwezuva rimwe nerimwe sevatengi.

Kufanotaura kwemamiriro ekunze kunogona kuita kushandiswa kwechikwata kuudza kana zuva richava kunaya, kupenya kana kuti kunaya. Chiremba inogona kuongorora mararamiro ehutano kufanotaura zvido zvekurapa. Mhando yekugadzirisa nzira, Naive Bayesian, inoshandisa mikana yepamusoro yekuisa zvikamu zve spam emails. Kubva pakunyengedzwa kwekunyengedzera kune zvigadzirwa zvinogadzirwa, kushambadzira kunotevera kumashure zuva rega rega kuongorora dambudziko uye kubudisa maitiro.