UDmitry Kazakov, Umholi Wethimba Lokuhlaziya Idatha ku-Kolesa Group, wabelana ngemininingwane evela kunhlolovo yokuqala yase-Kazakhstan yochwepheshe bedatha.
Esithombeni: Dmitry Kazakov
Khumbula umusho odumile othi i-Big Data ifana kakhulu nocansi lwentsha - wonke umuntu ukhuluma ngayo, kodwa akekho owaziyo ukuthi lukhona ngempela yini. Okufanayo kungashiwo mayelana nemakethe yochwepheshe bedatha (eKazakhstan) - kukhona i-hype, kodwa ubani ngemuva kwayo (nokuthi ngabe kukhona noma ubani lapho) kwakungacacile ngokuphelele - noma ku-HR, noma kubaphathi, noma ososayensi bedatha ngokwabo.
Sichithe , lapho bahlole khona ochwepheshe abangaphezu kuka-300 mayelana namaholo abo, imisebenzi, amakhono, amathuluzi nokunye okuningi.
I-Spoiler: Yebo, zikhona ngempela, kodwa konke akulula kangako.
Ukuqonda okuhle. Okokuqala, kunososayensi bedatha abaningi kunalokho ebesikulindele. Sikwazile ukuxoxisana nabantu abangu-300, phakathi kwabo okwakungebona abahlaziyi bemikhiqizo kuphela, ukumaketha kanye ne-BI, kodwa nonjiniyela be-ML kanye ne-DWH, okwakujabulisa kakhulu. Iqembu elikhulu kunawo wonke lalihlanganisa bonke labo abazibiza ngososayensi bedatha - lokho kungamaphesenti angama-36 abaphendulile. Kunzima ukusho ukuthi lokhu kuhlanganisa isidingo semakethe noma cha, ngoba imakethe ngokwayo iyakhiwa.
Ukusatshalaliswa kwamazinga emisebenzi kuyadida - cishe baningi abaholi bamaqembu nabaphathi njengabancane. Kungase kube nezizathu eziningana zalokhu. Isibonelo, inani elikhulu lamaqembu amancane abantu abangu-2-3, lapho umholi angaba uchwepheshe wezinga eliphakathi noma eliphezulu.
Esinye isizathu kungase kube isiphithiphithi esibusayo njengamanje emakethe mayelana namazinga ekusabalaliseni izindima nokusebenza. Abaholi bethimba ngezinye izikhathi babelwa labo abavele basebenze unyaka noma emibili isikhathi eside kunabanye, ngaphandle kokubhekisela ezingeni lamakhono nolwazi. Lokhu sikubona ekusatshalalisweni kwemisebenzi ngezikhundla - u-38% wabaphathi nabaholi bamaqembu benza umsebenzi wokucubungula ngaphambilini kanti omunye u-33% uhlaziya izibalo eziyisisekelo.
Lapha sicele abaphendulayo ukuthi bahlole ngokuzimele izinga lezibalo ezinkampanini zabo. Uma ubhekisisa, ungabona ukuthi u-10% wabaphenduli abasebenza eminyangweni yezibalo yabantu abangu-2-3 bakholelwa ukuthi "banezinga elithuthukisiwe."
Liyini “izinga elithuthukisiwe”? Uhlelo lwe-BI lusebenza kahle. Kukhona i-DWH kanye ne-Big Data. Ukuhlolwa kwe-A/B kwenziwa njalo. Kunezinhlelo ezisebenzayo ze-ML ne-DS ekukhiqizeni. Izinqumo zenziwa ngokusekelwe kudatha kuphela. Umnyango wokucutshungulwa kwedatha kanye nesayensi yedatha ungomunye obalulekile enkampanini.
Cishe akunakwenzeka ukufeza konke lokhu okungenhla ngomnyango wabantu abangu-2-3. Ngicabanga ukuthi lo mphumela wenhlolovo uyizinhlungu ezikhula kancane - abafana abakabi namuntu abangaziqhathanisa naye ukuze banqume izinga labo ngokufanele.
Njengoba kulindelekile, ososayensi bedatha abachitha isikhathi sabo esiningi hhayi kwizibalo eziyinkimbinkimbi kakhulu noma ubunjiniyela, kodwa ekucubunguleni kwangaphambili, ukulanda, nokuhlanza idatha. Kuwo wonke amakhono sibona ukucubungula kwangaphambili koku-3 okuphezulu. Kodwa asivamile ukubona izinto eziyinkimbinkimbi njengokwenza amamodeli e-ML noma ukusebenza ne-Big Data koku-3 okuphezulu - phakathi konjiniyela be-ML ne-DWH kuphela.
Kukhona nemibono embalwa edabukisayo. Ochwepheshe bazibekela ama-40% emisebenzi yabo ngokwabo. E-Kazakhstan, kuze kube manje kuphela izinkampani eziphezulu ze-unicorn eziye zazama izinzuzo zokusebenza ngedatha enkulu futhi zafunda ukuthi zingayenza kanjani ngendlela efanele. Basakaza emakethe ukuthi i-Big Data kanye ne-Machine Learning ipholile, futhi i-echelon yesibili ilandela ngemuva, kodwa ayiqondi ngaso sonke isikhathi ukuthi ukusebenza ngedatha kusebenza kanjani. Ngakho-ke, siyabona ukuthi ochwepheshe bazibekela imisebenzi, futhi amabhizinisi awazi njalo ukuthi afunani.
Kungimangazile ukuthi u-20% wongoti abazi noma inkampani yabo ineData Warehouse. Yebo, futhi ngezinhlelo zokuphatha isizindalwazi akuyona yonke into enhle kakhulu - ama-41% asebenzisa i-MySQL, kanti amanye ama-34% asebenzisa i-PostgreSQL. Lokhu kungasho ukuthini? Basebenza kunalokho ngedatha encane.
Embuzweni mayelana nezinhlelo zokugcina, siphinde sibone i-MySQL kanye (!) I-Excel. Kodwa lokhu kungase kubonise, isibonelo, ukuthi izinkampani eziningi azikabi naso isicelo sokusebenza ngedatha enkulu.
Lapha yonke into isiphinde yaba yindida. Ngokuvamile, amaholo ayengaphansi kancane kunalokho engangikulindele.
Ngokwami, kunzima kimi ukucabanga ngonjiniyela we-ML olungele ukusebenzela i-tenge eyizinkulungwane ezingama-200 - kungenzeka ukuthi ungumfundi. Kungakhathaliseki ukuthi amakhono ochwepheshe abanjalo abuthakathaka kakhulu, noma kusenzima ezinkampanini ukuhlola ngokwanele umsebenzi we-Data Science. Kodwa mhlawumbe lokhu kuphinde kubonise ukuthi imakethe isesekuqaleni kokuvuthwa kwayo. Futhi ngokuhamba kwesikhathi, izinga lamaholo lizosungulwa ezingeni elanele.
Source: www.habr.com