Big Data matekinoroji ave kushandiswa kwese kwese - muindasitiri, mushonga, bhizinesi, uye varaidzo. Nokudaro, pasina kuongorora deta huru, vatengesi vakuru havazokwanisi kushanda zvakanaka, kutengesa kuAmazon kuchawa, uye meteorologists havazokwanisi kufanotaura mamiriro ekunze kwemazuva mazhinji, mavhiki nemwedzi zvisati zvaitika. Zvine musoro kuti nyanzvi dzedata hombe dzava kudiwa zvakanyanya, uye kudiwa kuri kuramba kuchikura.
GeekBrains inodzidzisa vamiriri veiyi ndima, vachiedza kupa vadzidzi ruzivo rwedzidziso uye dzidziso nemienzaniso, iyo ine ruzivo nyanzvi inobatanidzwa. Gore rino Big Data vaongorori kubva paIndaneti yunivhesiti GeekUniversity uye mukuru mutengesi muRussia Federation, X5 Retail Group, vave vadyidzani. Nyanzvi dzekambani iyi, dzine ruzivo rwakakura uye ruzivo, dzakabatsira kugadzira kosi yemhando, umo vadzidzi vanogashira zvese dzidziso yedzidziso uye ruzivo runoshanda panguva yekudzidziswa.
Takataura naValery Babushkin, director of modelling uye data analysis paX5 Retail Group. Ndiye mumwe we data masayendisiti munyika (yemakumi matatu mune yepasi rose chinzvimbo chemuchina kudzidza nyanzvi). Pamwe chete nevamwe vadzidzisi, Valery anoudza vadzidzi veGeekBrains nezvekuyedzwa kweA/B, huwandu hwemasvomhu pakavakirwa nzira idzi, pamwe nemaitiro echizvino-zvino ekuverenga uye maficha ekuita bvunzo dzeA/B mukutengesa kunze kwenyika.
Nei tichida bvunzo dzeA/B zvachose?
Iyi ndiyo imwe yedzakanakisa nzira dzekutsvaga nzira dzakanakisa dzekuvandudza shanduko, hupfumi uye maitiro. Kune dzimwe nzira, asi dzinodhura uye dzakaoma. Iwo mabhenefiti makuru eA/B bvunzo mutengo wavo wakaderera uye kuwanikwa kwemabhizinesi chero saizi.
Nezve bvunzo dzeA / B, tinogona kutaura kuti iyi ndiyo imwe yedzakanyanya kukosha nzira dzekutsvaga nekuita sarudzo mubhizinesi, sarudzo idzo purofiti uye kuvandudzwa kwezvakasiyana zvigadzirwa zvechero kambani zvinoenderana. Miedzo inoita kuti zvikwanise kuita sarudzo zvichibva kwete chete padzidziso uye fungidziro, asiwo paruzivo runoshanda rwekuti shanduko chaiyo inoshandura sei kudyidzana kwevatengi netiweki.
Zvakakosha kuyeuka kuti mukutengesa unofanira kuedza zvose - mishandirapamwe yekushambadzira, kutumira mameseji eSMS, bvunzo dzekutumira ivo pachavo, kuiswa kwezvigadzirwa pamasherufu uye masherufu pachawo munzvimbo dzekutengesa. Kana tikataura nezvechitoro chendaneti, saka pano unogona kuyedza kurongeka kwezvinhu, dhizaini, zvinyorwa uye zvinyorwa.
A/B bvunzo chishandiso chinobatsira kambani, semuenzaniso, mutengesi, kuti agare achikwikwidza, pfungwa inoshanduka munguva uye ichinje pachayo. Izvi zvinobvumira bhizinesi kuti riite zvakanaka sezvinobvira, kuwedzera purofiti.
Ndedzipi nuances dzenzira idzi?
Chinhu chikuru ndechekuti panofanirwa kuve nechinangwa kana dambudziko pane iyo yekuedzwa ichave yakavakirwa. Semuenzaniso, dambudziko ndere nhamba diki yevatengi pane imwe nzvimbo yekutengesa kana online chitoro. Chinangwa ndechekuwedzera kuwanda kwevatengi. Hypothesis: kana makadhi echigadzirwa muchitoro chepamhepo akagadzirwa akakura uye mafoto akajeka, ipapo pachave nekutenga kwakawanda. Tevere, bvunzo yeA/B inoitwa, mhedzisiro yacho kuongororwa kwekuchinja. Mushure mekunge mhedzisiro yebvunzo dzese dzagamuchirwa, unogona kutanga kugadzira hurongwa hwekuchinja saiti.
Hazvikurudzirwe kuita bvunzo nemaitiro anopindirana, zvikasadaro mhedzisiro ichanyanya kuoma kuongorora. Zvinokurudzirwa kuita bvunzo pazvibodzwa zvekutanga uye kugadzira hypotheses kutanga.
Muedzo unofanirwa kugara kwenguva yakakwana kuti mhedzisiro yacho ionekwe seyakavimbika. Zvakawanda zvakadii zvinoenderana, hongu, pabvunzo pachayo. Saka, paZuva reGore Idzva, traffic yezvitoro zvakawanda zvepamhepo inowedzera. Kana dhizaini yechitoro chendaneti yakashandurwa kare, ipapo bvunzo yenguva pfupi icharatidza kuti zvese zvakanaka, shanduko dzinobudirira, uye traffic iri kukura. Asi kwete, zvisinei kuti chii chaunoita pamberi pezororo, traffic ichawedzera, bvunzo haigone kupedzwa pamberi peGore Idzva kana pakarepo mushure mayo, inofanira kunge yakareba zvakakwana kuti uone kuwirirana kwese.
Kukosha kwekubatana kwakarurama pakati pechinangwa uye chiratidzo chiri kuyerwa. Semuenzaniso, nekushandura dhizaini yewebhusaiti yechitoro chendaneti, kambani inoona kuwedzera kwenhamba yevashanyi kana vatengi uye inogutsikana neizvi. Asi kutaura zvazviri, avhareji cheki saizi inogona kunge idiki pane zvakajairwa, saka mari yako yese ichave yakaderera. Izvi, zvechokwadi, hazvigoni kunzi mugumisiro wakanaka. Dambudziko nderekuti kambani haina panguva imwechete kutarisa hukama pakati pekuwedzera kwevashanyi, kuwedzera kwenhamba yekutenga, uye masimba ehukuru hweavhareji cheki.
Kuedza kuzvitoro zvepa online chete here?
Kana bodo. Nzira ine mukurumbira mukutengesa kunze kwenyika ndeyekuitwa kwepombi yakazara yekuyedza hypotheses pasina Indaneti. Uku ndiko kuvakwa kwenzira umo njodzi dzekusarudzwa kwakashata kwemapoka ekuedza kunoderedzwa, chiyero chakanyanya chehuwandu hwezvitoro, nguva yekutyaira uye ukuru hwemhedzisiro inofungidzirwa inosarudzwa. Ndiko zvakare kushandiswa zvakare uye kuenderera mberi kwekuvandudza kwemashure-maitiro ekuongorora maitiro. Iyo nzira inodiwa kuderedza mukana wekukanganisa kwekubvuma kwenhema uye kukanganisa kukanganisa, pamwe nekuwedzera kunzwisisika, nokuti kunyange chidimbu chiduku pachiyero chebhizimisi guru chakakosha zvikuru. Naizvozvo, iwe unofanirwa kukwanisa kuona kunyangwe shanduko dzisina kusimba uye kuderedza njodzi, kusanganisira zvisizvo mhedziso nezvemhedzisiro yekuyedza.
Retail, Big Data uye chaiwo makesi
Gore rapfuura, nyanzvi dzeX5 Retail Group dzakaongorora mafambiro ekutengesa mavhoriyamu ezvigadzirwa zvakakurumbira pakati pevateveri ve2018 World Cup. Pakanga pasina zvinoshamisa, asi nhamba dzakaramba dzichinakidza.
Saka, mvura yakazova βNo. Mumaguta akatambira World Cup, kutengeswa kwemvura kwakawedzera neinenge 1%; mutungamiri aive Sochi, uko kuburitswa kwakawedzera ne46%. Pamazuva emutambo, nhamba yepamusoro yakanyorwa muSaransk - pano kutengesa kwakawedzera ne87% kana ichienzaniswa nemazuva ose.
Pamusoro pemvura, mafeni akatenga doro. Kubva munaChikumi 14 kusvika Chikunguru 15, mumaguta makaitika machisi, kutengesa kwedoro kwakawedzera neavhareji ye31,8%. Sochi akavawo mutungamiri - doro rakatengwa pano 64% zvakanyanya. Asi muSt. Petersburg kukura kwaiva kuduku - chete 5,6%. Pamazuva emitambo muSaransk, kutengesa doro kwakawedzera ne128%.
Tsvakurudzo yakaitwawo pane zvimwe zvigadzirwa. Data inowanikwa pamazuva epamusoro ekushandiswa kwechikafu inotibvumira kunyatsofanotaura kudiwa mune ramangwana, tichifunga nezvezviitiko zvezviitiko. Kufanotaura kwakarurama kunoita kuti zvikwanise kutarisira zvinotarisirwa nevatengi.
Panguva yekuyedza, X5 Retail Group yakashandisa nzira mbiri:
Bayesian structural time series modhi ine cumulative musiyano wefungidziro;
Regression ongororo nekuongorora kwekuchinja mukugovera kukanganisa pamberi uye panguva yemakwikwi.
Chii chimwe chinoshandiswa nekutengesa kubva kuBig Data?
- Kune akawanda akawanda maitiro uye matekinoroji, kubva pane anogona kunzi offhand, aya ndeaya:
- Demand forecast;
- Optimization yeassortment matrix;
- Chiono chekombuta yekuona voids pamasherufu uye kuona mutsara uchigadzira;
- Promo forecast.
Kushaikwa kwenyanzvi
Iko kudiwa kweBig Data nyanzvi kuri kuramba kuchikura. Nekudaro, muna 2018, nhamba yezvinzvimbo zvine chekuita nedata hombe yakawedzera kanomwe kana zvichienzaniswa ne7. Muhafu yekutanga ye2015, kudiwa kwenyanzvi kwakapfuura makumi matanhatu neshanu muzana ekudiwa kwegore rose ra2019.
Makambani makuru ari kunyanya kuda masevhisi eBig Data analyst. Semuyenzaniso, paMail.ru Group ivo vanodiwa mune chero purojekiti iyo data data, zvinyorwa zvemultimedia zvinogadziriswa, synthesis yekutaura uye kuongororwa kunoitwa (ichi, chekutanga, Cloud services, social network, mitambo, nezvimwewo). Huwandu hwevashandi mukambani iyi hwakapetwa katatu mumakore maviri apfuura. Mumwedzi misere yekutanga yegore rino, Mail.ru yakahaya nhamba yakafanana yeBig Data nyanzvi segore rose rapfuura. PaOzon, iyo Dhipatimendi reSainzi yakakura zvakapetwa katatu mumakore maviri apfuura. Mamiriro acho akafanana paMegafon - timu inoongorora data yakakura kakawanda mumakore 2,5 apfuura.
Pasina kupokana, mune ramangwana kudiwa kwevamiriri vehunyanzvi zvine chekuita neBig Data kuchakura zvakanyanya. Saka kana iwe uchifarira munzvimbo ino, unofanirwa kuedza ruoko rwako.
Source: www.habr.com