{"id":98221,"date":"2020-10-25T02:42:25","date_gmt":"2020-10-25T00:42:25","guid":{"rendered":"https:\/\/prohoster.info\/blog\/administrirovanie\/kak-stat-uspeshnym-data-sajentistom-i-analitikom-dannyh"},"modified":"2020-11-18T00:58:46","modified_gmt":"2020-11-17T22:58:46","slug":"kak-stat-uspeshnym-data-sajentistom-i-analitikom-dannyh","status":"publish","type":"post","link":"https:\/\/prohoster.info\/sq\/blog\/administrirovanie\/kak-stat-uspeshnym-data-sajentistom-i-analitikom-dannyh","title":{"rendered":"Si t\u00eb b\u00ebhesh nj\u00eb data scientist dhe analist t\u00eb t\u00eb dh\u00ebnave t\u00eb suksessh\u00ebm","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"

\"Si<\/a><\/noindex>
\nKa shum\u00eb artikuj p\u00ebr aft\u00ebsit\u00eb q\u00eb duhen p\u00ebr t\u2019u b\u00ebr\u00eb nj\u00eb data scientist ose analist i t\u00eb dh\u00ebnave i mir\u00eb, por vet\u00ebm pak prej tyre flasin p\u00ebr aft\u00ebsit\u00eb q\u00eb duhen p\u00ebr t\u00eb pasur sukses real \u2014 qoft\u00eb nj\u00eb vler\u00ebsim i shk\u00eblqyer i performanc\u00ebs, vler\u00ebsim nga drejtuesit, promovim apo t\u00eb gjitha bashk\u00eb. Sot ju sjellim nj\u00eb material, autorja e t\u00eb cilit d\u00ebshiron t\u00eb ndaj\u00eb p\u00ebrvoj\u00ebn e saj personale si data scientist dhe analiste e t\u00eb dh\u00ebnave, si edhe m\u00ebsimet q\u00eb ka nxjerr\u00eb p\u00ebr t\u00eb arritur suksesin.
\n<\/a><\/noindex><\/p>\n

Pata fat: m\u2019u ofrua nj\u00eb pozicion si data scientist n\u00eb nj\u00eb koh\u00eb kur nuk kisha fare p\u00ebrvoj\u00eb n\u00eb Data Science. Si ia dola mban\u00eb \u00ebsht\u00eb nj\u00eb histori m\u00eb vete, por dua t\u00eb them se para se ta pranoja k\u00ebt\u00eb pun\u00eb, kisha vet\u00ebm nj\u00eb ide t\u00eb paqart\u00eb p\u00ebr at\u00eb q\u00eb b\u00ebn nj\u00eb data scientist. <\/p>\n

M\u00eb pun\u00ebsuan p\u00ebr t\u00eb nd\u00ebrtuar data pipeline fal\u00eb p\u00ebrvoj\u00ebs sime t\u00eb m\u00ebparshme si inxhiniere e p\u00ebrpunimit t\u00eb t\u00eb dh\u00ebnave, ku zhvillova nj\u00eb data mart p\u00ebr analiz\u00eb parashikuese, q\u00eb p\u00ebrdorej nga ekipi i data scientist\u00ebve. <\/p>\n

Viti im i par\u00eb si data scientist p\u00ebrfshinte nd\u00ebrtimin e data pipeline p\u00ebr trajnimin e modeleve t\u00eb machine learning dhe vendosjen e tyre n\u00eb prodhim. Q\u00ebndroja n\u00eb prapasken\u00eb dhe nuk merrja pjes\u00eb n\u00eb shum\u00eb takime me pal\u00ebt e interesuara n\u00eb marketing, t\u00eb cilat ishin p\u00ebrdoruesit p\u00ebrfundimtar\u00eb t\u00eb modeleve. <\/p>\n

N\u00eb vitin tim t\u00eb dyt\u00eb n\u00eb kompani, u largua menaxheri i p\u00ebrpunimit dhe analiz\u00ebs s\u00eb t\u00eb dh\u00ebnave q\u00eb mbulonte marketingun. Q\u00eb nga ai moment, mora nj\u00eb rol qendror dhe u p\u00ebrfshiva m\u00eb aktivisht n\u00eb zhvillimin e modeleve dhe n\u00eb diskutimet p\u00ebr afatet e zbatimit t\u00eb projekteve.<\/p>\n

Gjat\u00eb komunikimit me pal\u00ebt e interesuara, kuptova se Data Science \u00ebsht\u00eb nj\u00eb koncept i paqart\u00eb, p\u00ebr t\u00eb cilin njer\u00ebzit kan\u00eb d\u00ebgjuar, por q\u00eb nuk e kuptojn\u00eb plot\u00ebsisht, sidomos drejtuesit e lart\u00eb.<\/p><\/blockquote>\n

\nKam nd\u00ebrtuar m\u00eb shum\u00eb se nj\u00ebqind modele, por vet\u00ebm nj\u00eb e treta e tyre u p\u00ebrdor, sepse nuk dija si t\u00eb tregoja vler\u00ebn e tyre, edhe pse modelet ishin k\u00ebrkuar fillimisht pik\u00ebrisht nga marketingu. <\/p>\n

Nj\u00eb nga an\u00ebtar\u00ebt e ekipit tim kaloi muaj t\u00eb t\u00ebr\u00eb duke zhvilluar nj\u00eb model q\u00eb, sipas drejtuesve t\u00eb lart\u00eb, do t\u00eb tregonte vler\u00ebn e ekipit t\u00eb data scientist-\u00ebve. Ideja ishte q\u00eb, pasi modeli t\u00eb p\u00ebrfundonte, ai t\u00eb shtrihej n\u00eb t\u00eb gjith\u00eb organizat\u00ebn dhe ekipet e marketingut t\u00eb inkurajoheshin ta p\u00ebrdornin.<\/p>\n

Kjo p\u00ebrfundoi n\u00eb nj\u00eb d\u00ebshtim t\u00eb plot\u00eb, sepse askush nuk e kuptonte se \u00e7far\u00eb ishte nj\u00eb model i machine learning dhe nuk arrinte t\u00eb kuptonte vler\u00ebn e p\u00ebrdorimit t\u00eb tij. N\u00eb fund, muaj t\u00eb t\u00ebr\u00eb u shpenzuan kot p\u00ebr di\u00e7ka q\u00eb askush nuk e donte.<\/p><\/blockquote>\n

\nNga situata t\u00eb tilla kam nxjerr\u00eb disa m\u00ebsime, t\u00eb cilat po i paraqes m\u00eb posht\u00eb.<\/p>\n

M\u00ebsimet q\u00eb kam nxjerr\u00eb p\u00ebr t'u b\u00ebr\u00eb nj\u00eb data scientist i suksessh\u00ebm<\/h2>\n

\n1. Vendoseni veten n\u00eb pozita suksesi duke zgjedhur kompanin\u00eb e duhur.<\/b>
\nGjat\u00eb intervist\u00ebs me kompanin\u00eb, pyesni p\u00ebr kultur\u00ebn e t\u00eb dh\u00ebnave dhe sa modele t\u00eb machine learning miratohen dhe p\u00ebrdoren n\u00eb vendimmarrje. K\u00ebrkoni shembuj konkret\u00eb. Zbuloni n\u00ebse infrastruktura e t\u00eb dh\u00ebnave \u00ebsht\u00eb gati p\u00ebr t\u00eb nisur modelimin. N\u00ebse 90% t\u00eb koh\u00ebs suaj e shpenzoni duke nxjerr\u00eb t\u00eb dh\u00ebna t\u00eb pap\u00ebrpunuara dhe duke i pastruar, pothuajse nuk do t'ju mbetet koh\u00eb p\u00ebr t\u00eb nd\u00ebrtuar modele q\u00eb t\u00eb tregojn\u00eb vler\u00ebn tuaj si data scientist. Tregoni kujdes n\u00ebse po pun\u00ebsoheni si data scientist p\u00ebr her\u00eb t\u00eb par\u00eb n\u00eb at\u00eb kompani. Kjo mund t\u00eb jet\u00eb si pozitive, ashtu edhe negative, n\u00eb var\u00ebsi t\u00eb kultur\u00ebs s\u00eb t\u00eb dh\u00ebnave. Mund t\u00eb p\u00ebrballeni me shum\u00eb rezistenc\u00eb gjat\u00eb zbatimit t\u00eb nj\u00eb modeli, n\u00ebse drejtuesit e lart\u00eb pun\u00ebsojn\u00eb nj\u00eb specialist t\u00eb Data Science vet\u00ebm sepse kompania d\u00ebshiron t\u00eb njihet si q\u00eb p\u00ebrdor Data Science p\u00ebr t\u00eb marr\u00eb vendime m\u00eb t\u00eb mira<\/i>, por n\u00eb t\u00eb v\u00ebrtet\u00eb nuk e ka iden\u00eb se \u00e7far\u00eb do t\u00eb thot\u00eb kjo. Gjithashtu, n\u00ebse gjeni nj\u00eb kompani q\u00eb drejtohet nga t\u00eb dh\u00ebnat (data driven), do t\u00eb rriteni bashk\u00eb me t\u00eb.<\/p>\n

2. Njihni t\u00eb dh\u00ebnat dhe treguesit ky\u00e7 t\u00eb performanc\u00ebs (KPI).<\/b>
\nN\u00eb fillim p\u00ebrmenda se, kur isha inxhiniere e p\u00ebrpunimit t\u00eb t\u00eb dh\u00ebnave, krijova nj\u00eb vitrin\u00eb t\u00eb t\u00eb dh\u00ebnave analitike p\u00ebr ekipin e data scientist-\u00ebve. Pasi u b\u00ebra vet\u00eb data scientist, arrita t\u00eb gjej mund\u00ebsi t\u00eb reja q\u00eb rrit\u00ebn sakt\u00ebsin\u00eb e modeleve, sepse n\u00eb rolin tim t\u00eb m\u00ebparsh\u00ebm kisha punuar intensivisht me t\u00eb dh\u00ebna t\u00eb pap\u00ebrpunuara. <\/p>\n

Duke paraqitur rezultatet e nj\u00eb prej fushatave tona, arrita t\u00eb demonstroj modele q\u00eb gjeneronin norma m\u00eb t\u00eb larta konvertimi (n\u00eb p\u00ebrqindje), pas s\u00eb cil\u00ebs u mat nj\u00eb nga KPI-t\u00eb e fushat\u00ebs. Kjo tregoi vler\u00ebn e modelit p\u00ebr efikasitetin e biznesit, me t\u00eb cilin mund t\u00eb lidhet marketingu.<\/p>\n

3. Siguroni pranimin e modelit duke treguar vler\u00ebn e tij p\u00ebr pal\u00ebt e interesuara<\/b>
\nNuk do t\u00eb keni kurr\u00eb sukses si specialist i Data Science n\u00ebse pal\u00ebt e interesuara nuk i p\u00ebrdorin kurr\u00eb modelet tuaja p\u00ebr t\u00eb marr\u00eb vendime biznesi. Nj\u00eb m\u00ebnyr\u00eb p\u00ebr t\u00eb siguruar pranimin e modelit \u00ebsht\u00eb t\u00eb gjeni nj\u00eb pik\u00eb t\u00eb dhimbshme t\u00eb biznesit dhe t\u00eb tregoni se si modeli mund t\u00eb ndihmoj\u00eb. <\/p>\n

Pas nj\u00eb bisede me departamentin ton\u00eb t\u00eb shitjeve, kuptova se dy p\u00ebrfaq\u00ebsues punonin me koh\u00eb t\u00eb plot\u00eb duke shqyrtuar manualisht miliona p\u00ebrdorues n\u00eb baz\u00ebn e t\u00eb dh\u00ebnave t\u00eb kompanis\u00eb p\u00ebr t\u00eb identifikuar p\u00ebrdoruesit me licenca individuale q\u00eb kishin m\u00eb shum\u00eb gjasa t\u00eb kalonin n\u00eb licenca ekipi. P\u00ebr p\u00ebrzgjedhjen p\u00ebrdorej nj\u00eb grup kriteresh, por procesi k\u00ebrkonte shum\u00eb koh\u00eb, sepse p\u00ebrfaq\u00ebsuesit i shqyrtonin p\u00ebrdoruesit nj\u00eb nga nj\u00eb. Me ndihm\u00ebn e modelit q\u00eb zhvillova, p\u00ebrfaq\u00ebsuesit mund\u00ebn t\u00eb zgjidhnin p\u00ebrdoruesit me probabilitetin m\u00eb t\u00eb lart\u00eb p\u00ebr t\u00eb bler\u00eb nj\u00eb licenc\u00eb ekipi dhe t\u00eb rrisnin mund\u00ebsin\u00eb e konvertimit n\u00eb m\u00eb pak koh\u00eb. Kjo \u00e7oi n\u00eb nj\u00eb p\u00ebrdorim m\u00eb efikas t\u00eb koh\u00ebs duke rritur norm\u00ebn e konvertimit p\u00ebr KPI-t\u00eb kryesore me t\u00eb cilat mund t\u00eb lidhet departamenti i shitjeve.<\/p>\n

Kaluan disa vite dhe e gjeta veten duke zhvilluar vazhdimisht t\u00eb nj\u00ebjtat modele, nd\u00ebrsa ndihesha sikur nuk po m\u00ebsoja m\u00eb asgj\u00eb t\u00eb re. Vendosa t\u00eb k\u00ebrkoja nj\u00eb rol tjet\u00ebr dhe p\u00ebrfundimisht mora nj\u00eb pozicion si analiste e t\u00eb dh\u00ebnave. Dallimi n\u00eb p\u00ebrgjegj\u00ebsi ishte shum\u00eb i madh krahasuar me koh\u00ebn kur isha data scientist, edhe pse s\u00ebrish mb\u00ebshtesja marketingun.<\/p>\n

Ishte hera e par\u00eb q\u00eb analizova eksperimente A\/B dhe zbulova t\u00eb gjith\u00eb<\/i> m\u00ebnyrat se si nj\u00eb eksperiment mund t\u00eb d\u00ebshtoj\u00eb. Si data scientist, un\u00eb nuk punoja fare me A\/B testimin, sepse kjo ishte p\u00ebrgjegj\u00ebsi e ekipit t\u00eb eksperimenteve. Punoja n\u00eb nj\u00eb gam\u00eb t\u00eb gjer\u00eb analizash mbi t\u00eb cilat marketingu kishte ndikim \u2014 nga rritja e norm\u00ebs s\u00eb konvertimit p\u00ebr segmentin premium te angazhimi i p\u00ebrdoruesve dhe parandalimi i largimit t\u00eb tyre. M\u00ebsova shum\u00eb m\u00ebnyra t\u00eb ndryshme p\u00ebr t\u00eb analizuar t\u00eb dh\u00ebnat dhe kalova shum\u00eb koh\u00eb duke p\u00ebrmbledhur rezultatet, duke i prezantuar ato pal\u00ebve t\u00eb interesuara dhe drejtuesve t\u00eb lart\u00eb. Si data scientist, kryesisht punoja me nj\u00eb lloj modeli dhe rrall\u00eb mbaja prezantime. Le t\u00eb kalojm\u00eb disa vite p\u00ebrpara dhe t\u00eb shohim aft\u00ebsit\u00eb q\u00eb m\u00ebsova p\u00ebr t'u b\u00ebr\u00eb nj\u00eb analiste e suksesshme.<\/p>\n

Aft\u00ebsit\u00eb q\u00eb m\u00ebsova p\u00ebr t'u b\u00ebr\u00eb nj\u00eb analiste e suksesshme e t\u00eb dh\u00ebnave<\/h2>\n

\n1. M\u00ebsoni t\u00eb tregoni histori p\u00ebrmes t\u00eb dh\u00ebnave<\/b>
\nMos i shikoni KPI-t\u00eb t\u00eb ndara ve\u00e7mas. Lidhini ato mes tyre dhe shiheni biznesin si nj\u00eb t\u00ebr\u00ebsi. Kjo ju lejon t\u00eb identifikoni fushat q\u00eb ndikojn\u00eb nj\u00ebra-tjetr\u00ebn. Drejtuesit e lart\u00eb e shohin biznesin p\u00ebrmes k\u00ebsaj lenteje, dhe nj\u00eb person q\u00eb demonstron k\u00ebt\u00eb aft\u00ebsi bie n\u00eb sy kur vjen koha p\u00ebr t\u00eb marr\u00eb nj\u00eb vendim p\u00ebr promovim.<\/p>\n

2. Jepni ide t\u00eb zbatueshme<\/b>
\nJepini biznesit nj\u00eb ide t\u00eb zbatueshme<\/i> p\u00ebr t\u00eb zgjidhur problemin. Akoma m\u00eb mir\u00eb n\u00ebse propozoni zgjidhjen n\u00eb m\u00ebnyr\u00eb proaktive, edhe kur ende nuk \u00ebsht\u00eb th\u00ebn\u00eb se po p\u00ebrballeni me nj\u00eb problem prioritar. <\/p>\n

P\u00ebr shembull, n\u00ebse do t'i thoshit ekipit t\u00eb marketingut: \u00abVura re se koh\u00ebt e fundit numri i vizitor\u00ebve t\u00eb faqes po bie \u00e7do muaj\u00bb<\/i>. Kjo \u00ebsht\u00eb nj\u00eb prirje q\u00eb ata mund ta ken\u00eb v\u00ebn\u00eb re n\u00eb dashboard, dhe si analist nuk keni ofruar asnj\u00eb zgjidhje me vler\u00eb, sepse thjesht keni deklaruar nj\u00eb v\u00ebzhgim. <\/p>\n

N\u00eb vend t\u00eb k\u00ebsaj, analizoni t\u00eb dh\u00ebnat p\u00ebr t\u00eb gjetur shkakun dhe propozoni nj\u00eb zgjidhje. Nj\u00eb shembull m\u00eb i mir\u00eb p\u00ebr marketingun do t\u00eb ishte: \u00abVura re se koh\u00ebt e fundit na ka r\u00ebn\u00eb numri i vizitor\u00ebve n\u00eb faqen ton\u00eb t\u00eb internetit. Zbulova se burimi i problemit \u00ebsht\u00eb k\u00ebrkimi organik, p\u00ebr shkak t\u00eb ndryshimeve t\u00eb fundit q\u00eb kan\u00eb sjell\u00eb r\u00ebnie t\u00eb renditjes son\u00eb n\u00eb k\u00ebrkimin e Google\u00bb<\/i>. Kjo qasje tregon se keni monitoruar KPI-t\u00eb e kompanis\u00eb, keni v\u00ebn\u00eb re ndryshimin, keni hetuar shkakun dhe keni propozuar nj\u00eb zgjidhje p\u00ebr problemin.<\/p>\n

3. B\u00ebhuni nj\u00eb k\u00ebshilltar i besuesh\u00ebm<\/b>
\nDuhet t\u00eb b\u00ebheni personi i par\u00eb te i cili pal\u00ebt tuaja t\u00eb interesit drejtohen p\u00ebr rekomandime ose pyetje lidhur me fush\u00ebn e pun\u00ebs q\u00eb mbuloni. Nuk ka rrug\u00eb t\u00eb shkurt\u00ebr, sepse duhet koh\u00eb p\u00ebr t\u2019i treguar k\u00ebto aft\u00ebsi n\u00eb praktik\u00eb. \u00c7el\u00ebsi \u00ebsht\u00eb t\u00eb ofroni vazhdimisht analiza cil\u00ebsore me sa m\u00eb pak gabime. \u00c7do gabim n\u00eb llogaritje do t\u2019ju kushtoj\u00eb n\u00eb besueshm\u00ebri, sepse her\u00ebn tjet\u00ebr kur t\u00eb paraqisni nj\u00eb analiz\u00eb, njer\u00ebzit mund t\u00eb pyesin veten: N\u00ebse keni gabuar her\u00ebn e kaluar, ndoshta po gaboni edhe k\u00ebsaj here?<\/i>. Gjithmon\u00eb kontrolloni dy her\u00eb pun\u00ebn tuaj. Gjithashtu, nuk b\u00ebn keq t\u2019i k\u00ebrkoni menaxherit ose nj\u00eb kolegu t\u00eb hedh\u00eb nj\u00eb sy mbi shifrat tuaja para se t\u2019i prezantoni, n\u00ebse keni ndonj\u00eb dyshim p\u00ebr analiz\u00ebn tuaj.<\/p>\n

4. M\u00ebsoni t\u00eb komunikoni qart\u00eb rezultate komplekse<\/b>
\nS\u00ebrish, nuk ka asnj\u00eb rrug\u00eb t\u00eb shkurt\u00ebr p\u00ebr t\u00eb m\u00ebsuar t\u00eb komunikoni me efektivitet. Kjo k\u00ebrkon praktik\u00eb dhe, me kalimin e koh\u00ebs, do t\u00eb b\u00ebheni m\u00eb t\u00eb zot\u00eb n\u00eb k\u00ebt\u00eb drejtim. Gj\u00ebja kryesore \u00ebsht\u00eb t\u00eb p\u00ebrcaktoni pikat thelb\u00ebsore t\u00eb asaj q\u00eb doni t\u00eb b\u00ebni dhe t\u00eb rekomandoni \u00e7do veprim q\u00eb pal\u00ebt e interesit mund t\u00eb nd\u00ebrmarrin, bazuar n\u00eb analiz\u00ebn tuaj, p\u00ebr t\u00eb p\u00ebrmir\u00ebsuar biznesin. Sa m\u00eb lart t\u00eb ngjiteni n\u00eb karrier\u00eb brenda organizat\u00ebs, aq m\u00eb e r\u00ebnd\u00ebsishme b\u00ebhet aft\u00ebsia p\u00ebr t\u00eb komunikuar. Paraqitja e rezultateve komplekse \u00ebsht\u00eb nj\u00eb aft\u00ebsi ky\u00e7e q\u00eb duhet ta demonstroni. Kam kaluar vite duke m\u00ebsuar sekretet e suksesit si Data Scientist dhe analiste e t\u00eb dh\u00ebnave. Njer\u00ebzit e p\u00ebrkufizojn\u00eb suksesin n\u00eb m\u00ebnyra t\u00eb ndryshme. P\u00ebr mua, suksesi \u00ebsht\u00eb t\u00eb t\u00eb quajn\u00eb nj\u00eb analist t\u00eb jasht\u00ebzakonsh\u00ebm dhe t\u00eb spikatur. Tani q\u00eb i njihni k\u00ebto sekrete, shpresoj q\u00eb rruga juaj drejt suksesit t\u00eb jet\u00eb m\u00eb e shpejt\u00eb, pavar\u00ebsisht se si e p\u00ebrkufizoni at\u00eb.<\/p>\n

Dhe q\u00eb rruga juaj drejt suksesit t\u00eb jet\u00eb edhe m\u00eb e shpejt\u00eb, ja ku keni nj\u00eb kod promocional HABR<\/b>, me t\u00eb cilin mund t\u00eb merrni 10% ulje shtes\u00eb mbi zbritjen e treguar n\u00eb baner.<\/p>\n

\"Si<\/a><\/noindex><\/p>\n