{"id":32907,"date":"2019-10-31T21:49:39","date_gmt":"2019-10-31T18:49:39","guid":{"rendered":"https:\/\/prohoster.info\/blog\/pochemu-data-science-komandam-nuzhny-universaly-a-ne-spetsialisty\/"},"modified":"2019-10-31T21:49:39","modified_gmt":"2019-10-31T18:49:39","slug":"pochemu-data-science-komandam-nuzhny-universaly-a-ne-spetsialisty","status":"publish","type":"post","link":"https:\/\/prohoster.info\/sk\/blog\/news\/pochemu-data-science-komandam-nuzhny-universaly-a-ne-spetsialisty","title":{"rendered":"Pre\u010do potrebuj\u00fa t\u00edmom Data Science univerz\u00e1lov, a nie \u0161pecialistov?","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"

\"Pre\u010do
\nHIROSHI WATANABE\/GETTY IMAGES<\/i><\/p>\n

V knihe \u201eBohatstvo n\u00e1rodov\u201c Adam Smith ukazuje, ako rozdelenie pr\u00e1ce sa st\u00e1va hlavn\u00fdm zdrojom zv\u00fd\u0161enia produktivity. Pr\u00edkladom je mont\u00e1\u017ena linka v tov\u00e1rni na v\u00fdrobu \u0161pendl\u00edkov: \u201eJeden pracovn\u00edk \u0165ah\u00e1 dr\u00f4t, druh\u00fd ho narovn\u00e1va, tret\u00ed ho strih\u00e1, \u0161tvrt\u00fd ostr\u00ed koniec, piaty obr\u00fasi druh\u00fd koniec na nasadenie hlavi\u010dky\u201c. V\u010faka \u0161pecializ\u00e1cii zameranej na konkr\u00e9tne funkcie sa ka\u017ed\u00fd zamestnanec st\u00e1va vysoko kvalifikovan\u00fdm odborn\u00edkom vo svojej \u00fazkej \u00falohe, \u010do vedie k zv\u00fd\u0161eniu efektivity procesu. V\u00fdroba na jedn\u00e9ho pracovn\u00edka sa mnohon\u00e1sobne zvy\u0161uje a tov\u00e1re\u0148 sa st\u00e1va efekt\u00edvnej\u0161ou vo v\u00fdrobe \u0161pendl\u00edkov.<\/p>\n

Tak\u00e9to rozdelenie pr\u00e1ce pod\u013ea funk\u010dnosti je v na\u0161ich mysliach tak zakorenen\u00e9 aj dnes, \u017ee r\u00fdchlo organizujeme na\u0161e t\u00edmu zodpovedaj\u00facim sp\u00f4sobom. Data Science nie je v\u00fdnimkou. Komplexn\u00e9 algoritmick\u00e9 obchodn\u00e9 pr\u00edle\u017eitosti vy\u017eaduj\u00fa mno\u017estvo pracovn\u00fdch funkci\u00ed, a preto spolo\u010dnosti zvy\u010dajne vytv\u00e1raj\u00fa t\u00edmy odborn\u00edkov: v\u00fdskumn\u00edkov, in\u017einierov na anal\u00fdzu \u00fadajov, in\u017einierov strojov\u00e9ho u\u010denia, vedcov zaoberaj\u00facich sa pr\u00ed\u010dinno-n\u00e1sledkov\u00fdmi vz\u0165ahmi a tak \u010falej. Pr\u00e1ca odborn\u00edkov je koordinovan\u00e1 produktov\u00fdm mana\u017e\u00e9rom s prenosom funkci\u00ed sp\u00f4sobom, ktor\u00fd pripom\u00edna tov\u00e1re\u0148 na \u0161pendl\u00edky: \u201ejeden \u010dlovek z\u00edskava \u00fadaje, druh\u00fd ich modeluje, tret\u00ed realizuje, \u0161tvrt\u00fd meria\u201c a tak \u010falej.
\n<\/a><\/noindex>
\nBohu\u017eia\u013e, nemali by sme optimalizova\u0165 na\u0161e Data Science t\u00edmy na zv\u00fd\u0161enie produktivity. Av\u0161ak rob\u00edte to, ke\u010f pochop\u00edte, \u010do vyr\u00e1bate: \u0161pendl\u00edky alebo nie\u010do in\u00e9, a jednoducho sa usilujete o zv\u00fd\u0161enie efektivity. Cie\u013eom mont\u00e1\u017enych liniek je vykonanie \u00falohy. Presne vieme, \u010do chceme \u2014 s\u00fa to \u0161pendl\u00edky (ako v Smithovom pr\u00edklade), ale m\u00f4\u017eeme spomen\u00fa\u0165 ak\u00fdko\u013evek produkt alebo slu\u017ebu, kde po\u017eiadavky \u00faplne opisuj\u00fa v\u0161etky aspekty produktu a jeho spr\u00e1vanie. \u00dalohou zamestnancov je splni\u0165 tieto po\u017eiadavky \u010do najefekt\u00edvnej\u0161ie.<\/p>\n

Cie\u013eom Data Science nie je iba vykon\u00e1va\u0165 \u00falohy. Sk\u00f4r ide o sk\u00famanie a rozvoj nov\u00fdch siln\u00fdch obchodn\u00fdch mo\u017enost\u00ed. Algoritmick\u00e9 produkty a slu\u017eby, ako s\u00fa odpor\u00fa\u010dacie syst\u00e9my, z\u00e1kazn\u00edcke interakcie, klasifik\u00e1cia preferenci\u00ed v \u0161t\u00fdle, v\u00fdber ve\u013ekost\u00ed, dizajn oble\u010denia, optimaliz\u00e1cia logistiky, odha\u013eovanie sez\u00f3nnych trendov a mnoho \u010fal\u0161ieho, nem\u00f4\u017eu by\u0165 navrhnut\u00e9 vopred. Musia by\u0165 presk\u00faman\u00e9. Neexistuj\u00fa \u017eiadne v\u00fdkresy, ktor\u00e9 by ich reprodukovali; ide o nov\u00e9 mo\u017enosti s inherentnou neistotou. Koeficienty, modely, typy modelov, hyperparametre \u2013 v\u0161etky potrebn\u00e9 prvky musia by\u0165 presk\u00faman\u00e9 pomocou experimentov, pokusov a omylov, ako aj opakovania. Pri pinov\u00fdch technik\u00e1ch sa vzdel\u00e1vanie a dizajn vykon\u00e1vaj\u00fa vopred, pred ich v\u00fdrobou. S Data Science sa u\u010d\u00edte po\u010das procesu, nie pred n\u00edm.<\/p>\n

Na tov\u00e1rni na \u0161pendl\u00edky, kde je vzdel\u00e1vanie na prvom mieste, ne\u010dak\u00e1me ani nechceme, aby pracovn\u00edci improvizovali s akouko\u013evek vlastnos\u0165ou produktu, okrem zvy\u0161ovania \u00fa\u010dinnosti v\u00fdroby. \u0160pecializ\u00e1cia \u00faloh m\u00e1 zmysel, preto\u017ee vedie k efekt\u00edvnosti procesov a konzistentnosti v\u00fdroby (bez zmien v kone\u010dnom produkte).<\/p>\n

Ale ke\u010f sa produkt st\u00e1le vyv\u00edja a cie\u013e je vzdel\u00e1vanie, \u0161pecializ\u00e1cia n\u00e1m br\u00e1ni v dosahovan\u00ed na\u0161ich cie\u013eov v nasleduj\u00facich pr\u00edpadoch:<\/p>\n

1. Zvy\u0161uje n\u00e1klady na koordin\u00e1ciu. <\/b><\/p>\n

To znamen\u00e1 n\u00e1klady, ktor\u00e9 sa hromadia v priebehu \u010dasu na komunik\u00e1ciu, diskusiu, zd\u00f4vodnenie a stanovenie prior\u00edt pr\u00e1ce, ktor\u00fa je potrebn\u00e9 vykona\u0165. Tieto n\u00e1klady sa \u0161k\u00e1luj\u00fa nadlinne s po\u010dtom zapojen\u00fdch \u013eud\u00ed. (Ako n\u00e1s u\u010dil J. Richard Hackman, po\u010det vz\u0165ahov r rastie podobne ako funkcia po\u010dtu \u010dlenov n pod\u013ea tejto rovnice: r = (n ^ 2 - n) \/ 2. A ka\u017ed\u00fd vz\u0165ah odha\u013euje ur\u010dit\u00fa \u00farove\u0148 n\u00e1kladov\u00fdch s\u00favislost\u00ed). Ke\u010f s\u00fa odborn\u00edci na anal\u00fdzu d\u00e1t organizovan\u00ed pod\u013ea funkci\u00ed, v ka\u017edej f\u00e1ze, s ka\u017edou zmenou, ka\u017ed\u00fdm odovzdan\u00edm z\u00e1znamu a pod., je potrebn\u00fdch ve\u013ea odborn\u00edkov, \u010do zvy\u0161uje n\u00e1klady na koordin\u00e1ciu. Napr\u00edklad odborn\u00edci na \u0161tatistick\u00e9 modelovanie, ktor\u00ed sa chc\u00fa experiment\u00e1lne zaobera\u0165 nov\u00fdmi funkciami, bud\u00fa musie\u0165 koordinova\u0165 svoje kroky s in\u017einiermi pre spracovanie d\u00e1t, ktor\u00ed dop\u013a\u0148aj\u00fa d\u00e1tov\u00e9 s\u00fabory v\u017edy, ke\u010f chc\u00fa vysk\u00fa\u0161a\u0165 nie\u010do nov\u00e9. Rovnako ka\u017ed\u00e1 nov\u00e1 natr\u00e9novan\u00e1 model znamen\u00e1, \u017ee v\u00fdvoj\u00e1r modelu potrebuje niekoho, s k\u00fdm by mohol koordinova\u0165 svoje kroky pri jej uveden\u00ed do prev\u00e1dzky. N\u00e1klady na koordin\u00e1ciu p\u00f4sobia ako poplatok za iter\u00e1ciu, \u010do ich rob\u00ed n\u00e1ro\u010dnej\u0161\u00edmi a drah\u0161\u00edmi a s v\u00e4\u010d\u0161ou pravdepodobnos\u0165ou sp\u00f4sobuj\u00fa, \u017ee sa upust\u00ed od v\u00fdskumu. To m\u00f4\u017ee br\u00e1ni\u0165 vzdel\u00e1vaniu.<\/p>\n

2. To komplikuje \u010das \u010dakania. <\/b><\/p>\n

E\u0161te desivej\u0161ie ako n\u00e1klady na koordin\u00e1ciu je \u010das straten\u00fd medzi pracovn\u00fdmi zmenami. Zatia\u013e \u010do n\u00e1klady na koordin\u00e1ciu sa zvy\u010dajne meraj\u00fa v hodin\u00e1ch: \u010dase potrebnom na usporiadanie sch\u00f4dz\u00ed, diskusi\u00ed a presk\u00faman\u00ed projektov, \u010das \u010dakania sa zvy\u010dajne meria v d\u0148och, t\u00fd\u017ed\u0148och alebo dokonca mesiacoch! Harmonogramy funk\u010dn\u00fdch \u0161pecialistov je \u0165a\u017ek\u00e9 zos\u00faladi\u0165, preto\u017ee ka\u017ed\u00fd \u0161pecialista mus\u00ed by\u0165 prerozdelen\u00fd medzi nieko\u013eko projektov. Jednohodinov\u00e9 stretnutie na diskusiu o zmen\u00e1ch m\u00f4\u017ee trva\u0165 nieko\u013eko t\u00fd\u017ed\u0148ov, aby sa zos\u00faladil pracovn\u00fd tok. A po schv\u00e1len\u00ed zmien je potrebn\u00e9 napl\u00e1nova\u0165 aj samotn\u00fa skuto\u010dn\u00fa pr\u00e1cu v kontexte mno\u017estva in\u00fdch projektov, \u010do zaber\u00e1 pracovn\u00fd \u010das \u0161pecialistov. Pr\u00e1ca s\u00favisiaca s opravou k\u00f3du alebo prieskumu, na ktor\u00fa je potrebn\u00fdch iba nieko\u013eko hod\u00edn alebo dn\u00ed, m\u00f4\u017ee trva\u0165 ove\u013ea dlh\u0161ie, k\u00fdm sa zdroje stan\u00fa dostupn\u00fdmi. Do tej doby sa iter\u00e1cie a u\u010denie pozastavuj\u00fa.<\/p>\n

3. Z\u00fa\u017ei kontext.<\/b><\/p>\n

Rozdelenie pr\u00e1ce m\u00f4\u017ee umelo obmedzi\u0165 u\u010denie, odme\u0148uj\u00fac \u013eud\u00ed za to, \u017ee zost\u00e1vaj\u00fa vo svojej \u0161pecializ\u00e1cii. Napr\u00edklad v\u00fdskumn\u00fd vedec, ktor\u00fd mus\u00ed zosta\u0165 v r\u00e1mci svojej funk\u010dnosti, zameria svoju energiu na experimenty s r\u00f4znymi typmi algoritmov: regresiami, neur\u00f3nov\u00fdmi sie\u0165ami, random forest a tak \u010falej. Ur\u010dite dobr\u00fd v\u00fdber algoritmu m\u00f4\u017ee vies\u0165 k postupn\u00fdm zlep\u0161eniam, ale spravidla sa d\u00e1 z\u00edska\u0165 omnoho viac z in\u00fdch \u010dinnost\u00ed, ako je integr\u00e1cia nov\u00fdch zdrojov \u00fadajov. Rovnako to pom\u00f4\u017ee vyvin\u00fa\u0165 model, ktor\u00fd vyu\u017e\u00edva ka\u017ed\u00fa \u010das\u0165 vysvet\u013euj\u00facej schopnosti, ktor\u00fa maj\u00fa \u00fadaje. Napriek tomu m\u00f4\u017ee by\u0165 jej silnou str\u00e1nkou zmena cie\u013eovej funkcie alebo oslabenie ur\u010dit\u00fdch obmedzen\u00ed. Je \u0165a\u017ek\u00e9 to vidie\u0165 alebo urobi\u0165, ke\u010f je jej pr\u00e1ca obmedzen\u00e1. Preto\u017ee odborn\u00fd vedec sa \u0161pecializuje na optimaliz\u00e1ciu algoritmov, m\u00e1 ove\u013ea menej \u0161anc\u00ed zaobera\u0165 sa nie\u010d\u00edm in\u00fdm, aj ke\u010f to prin\u00e1\u0161a v\u00fdznamn\u00fd pr\u00ednos.<\/p>\n

Uve\u010fme znaky, ktor\u00e9 sa objavia, ke\u010f t\u00edmy v oblasti data science pracuj\u00fa ako tov\u00e1rne na \u0161pendl\u00edky (napr\u00edklad pri jednoduch\u00fdch aktualiz\u00e1ci\u00e1ch stavu): \u201e\u010dakanie na zmeny v toku d\u00e1t\u201c a \u201e\u010dakanie na zdroje ML Eng\u201c, ktor\u00e9 s\u00fa be\u017en\u00fdmi prek\u00e1\u017ekami. Av\u0161ak pova\u017eujem za nebezpe\u010dnej\u0161\u00ed vplyv to, \u010do si nev\u0161imnete, preto\u017ee nem\u00f4\u017eete \u013eutova\u0165 to, o \u010dom e\u0161te neviete. Bezchybn\u00e1 realiz\u00e1cia po\u017eiadaviek a sebauspokojenie, dosiahnut\u00e9 v d\u00f4sledku zefekt\u00edvnenia procesov, m\u00f4\u017eu zakr\u00fdva\u0165 pravdu o tom, \u017ee organiz\u00e1cie nie s\u00fa obozn\u00e1men\u00e9 s v\u00fdhodami u\u010denia, ktor\u00e9 prehliadaj\u00fa.<\/p>\n

Rie\u0161en\u00edm tohto probl\u00e9mu je samozrejme zbavi\u0165 sa tov\u00e1rensk\u00e9ho met\u00f3dy \u0161pendl\u00edka. Aby sme podporili u\u010denie a iter\u00e1ciu, \u00falohy v oblasti data science by mali by\u0165 spolo\u010dn\u00e9, ale s rozsiahlymi zodpovednos\u0165ami, ktor\u00e9 nie s\u00fa z\u00e1visl\u00e9 od technickej funkcie, to znamen\u00e1, organizova\u0165 odborn\u00edkov na d\u00e1ta tak, aby boli optimalizovan\u00ed na u\u010denie. To znamen\u00e1, \u017ee je potrebn\u00e9 zamestna\u0165 \u201e\u0161pecialistov na pln\u00fd stack\u201c \u2013 v\u0161eobecn\u00fdch odborn\u00edkov, ktor\u00ed m\u00f4\u017eu vykon\u00e1va\u0165 r\u00f4zne funkcie: od konceptu po modelovanie, od implement\u00e1cie po meranie. Je d\u00f4le\u017eit\u00e9 poznamena\u0165, \u017ee nepredpoklad\u00e1m, \u017ee pri naj\u00edman\u00ed \u0161pecialistov na pln\u00fd stack by sa mal zmen\u0161i\u0165 po\u010det zamestnancov. Sk\u00f4r si len predpoklad\u00e1m, \u017ee ke\u010f s\u00fa organizovan\u00ed inak, ich stimuly sa lep\u0161ie zhoduj\u00fa s v\u00fdhodami u\u010denia a efekt\u00edvnosti. Napr\u00edklad m\u00e1te t\u00edm troch \u013eud\u00ed, ktor\u00ed maj\u00fa tri obchodn\u00e9 zru\u010dnosti. V tov\u00e1rni na \u0161pendl\u00edky by sa ka\u017ed\u00fd odborn\u00edk venoval tretine \u010dasu ka\u017edej profesijnej \u00falohe, preto\u017ee nikto in\u00fd by nemohol splni\u0165 jeho pr\u00e1cu. V plnom stacku sa ka\u017ed\u00fd v\u0161estrann\u00fd zamestnanec plne anga\u017euje vo v\u0161etkom obchodnom procese, z\u00edskavan\u00ed objemu pr\u00e1ce a u\u010den\u00ed.<\/p>\n

S men\u0161\u00edm po\u010dtom \u013eud\u00ed podporuj\u00facich v\u00fdrobn\u00fd cyklus kles\u00e1 koordin\u00e1cia. Univerz\u00e1l plynulo prech\u00e1dza medzi funkciami, roz\u0161iruj\u00fac d\u00e1tov\u00fd kan\u00e1l, aby pridal v\u00e4\u010d\u0161\u00ed objem d\u00e1t, vysk\u00fa\u0161al nov\u00e9 funkcie v modeloch, nasadil nov\u00e9 verzie do v\u00fdroby pre kauz\u00e1lne merania a opakuje kroky tak r\u00fdchlo, ako prich\u00e1dzaj\u00fa nov\u00e9 n\u00e1pady. Samozrejme, univerz\u00e1l vykon\u00e1va r\u00f4zne funkcie sekven\u010dne, nie paralelne. Napokon, je to len jedna osoba. Av\u0161ak vykonanie \u00falohy zvy\u010dajne zaber\u00e1 len mal\u00fa \u010das\u0165 \u010dasu potrebn\u00e9ho na pr\u00edstup k in\u00e9mu \u0161pecializovan\u00e9mu zdroju. Tak\u017ee \u010das iter\u00e1cie sa zni\u017euje.<\/p>\n

N\u00e1\u0161 univerz\u00e1l mo\u017eno nie je tak zru\u010dn\u00fd ako \u0161pecialista v ur\u010ditej pracovnej funkcii, ale nesna\u017e\u00edme sa o funk\u010dn\u00e9 dokonalosti ani o mal\u00e9 postupn\u00e9 zlep\u0161enia. Namiesto toho sa sna\u017e\u00edme sk\u00fama\u0165 a objavova\u0165 v\u0161etky nov\u00e9 profesion\u00e1lne \u00falohy s postupn\u00fdm dopadom. S celostn\u00fdm kontextom pre komplexn\u00e9 rie\u0161enie vid\u00ed pr\u00edle\u017eitosti, ktor\u00e9 by preme\u0161kal \u00fazky \u0161pecialista. M\u00e1 viac n\u00e1padov a v\u00e4\u010d\u0161ie mo\u017enosti. Taktie\u017e zlyh\u00e1. Napriek tomu s\u00fa n\u00e1klady na zlyhanie mal\u00e9 a v\u00fdhody u\u010denia vysok\u00e9. T\u00e1to asymetria podporuje r\u00fdchlu iter\u00e1ciu a odme\u0148uje u\u010denie.<\/p>\n

Je d\u00f4le\u017eit\u00e9 poznamena\u0165, \u017ee tento rozsah auton\u00f3mie a rozmanitosti zru\u010dnost\u00ed poskytovan\u00fdch vedcom pracuj\u00facim s pln\u00fdmi stackmi do ve\u013ekej miery z\u00e1vis\u00ed od spo\u013eahlivosti d\u00e1tovej platformy, na ktorej sa d\u00e1 pracova\u0165. Dobre navrhnut\u00e1 d\u00e1tov\u00e1 platforma abstrahuje vedcov v oblasti spracovania d\u00e1t od zlo\u017eitost\u00ed kontajneriz\u00e1cie, distribuovan\u00e9ho spracovania, automatick\u00e9ho prechodu na in\u00fd zdroj a in\u00fdch pokro\u010dil\u00fdch po\u010d\u00edta\u010dov\u00fdch konceptov. Okrem abstrakcie m\u00f4\u017ee spo\u013eahliv\u00e1 d\u00e1tov\u00e1 platforma zabezpe\u010di\u0165 plynul\u00e9 pripojenie k experiment\u00e1lnej infra\u0161trukt\u00fare, automatizova\u0165 monitorovanie a syst\u00e9m upozornen\u00ed, zabezpe\u010di\u0165 automatick\u00e9 \u0161k\u00e1lovanie a vizualiz\u00e1ciu algoritmick\u00fdch v\u00fdsledkov a ladenia. Tieto komponenty s\u00fa navrhovan\u00e9 a vytv\u00e1ran\u00e9 in\u017einiermi d\u00e1tovej platformy, t.j. \u0161pecialista na data science ich nepred\u00e1va t\u00edmu v\u00fdvoj\u00e1rov d\u00e1tovej platformy. To \u0161pecialista na data science zodpoved\u00e1 za cel\u00fd k\u00f3d, ktor\u00fd sa pou\u017e\u00edva na spustenie platformy.<\/p>\n

Aj m\u0148a raz zaujala funk\u010dn\u00e1 de\u013eba pr\u00e1ce vyu\u017e\u00edvaj\u00faca efektivitu procesov, ale met\u00f3dou pokusov a omylov (lep\u0161\u00ed sp\u00f4sob u\u010denia neexistuje) som zistil, \u017ee typick\u00e9 \u00falohy lep\u0161ie prispievaj\u00fa k u\u010deniu a inov\u00e1ci\u00e1m a poskytuj\u00fa presn\u00e9 ukazovatele: objavovanie a budovanie ove\u013ea v\u00e4\u010d\u0161ieho po\u010dtu obchodn\u00fdch pr\u00edle\u017eitost\u00ed ne\u017e \u0161pecializovan\u00fd pr\u00edstup. (Efekt\u00edvnej\u0161\u00ed sp\u00f4sob, ako sa dozvedie\u0165 o tomto pr\u00edstupe k organiz\u00e1cii ako met\u00f3da pokusov a omylov, ktorou som pre\u0161iel, je pre\u010d\u00edta\u0165 si knihu Amy Edmondson \u201eInterakcia v t\u00edme: ako sa organiz\u00e1cie u\u010dia, vytv\u00e1raj\u00fa inov\u00e1cie a s\u00fa\u0165a\u017eia v ekonomike znalost\u00ed\u201c).<\/p>\n

Existuj\u00fa ur\u010dit\u00e9 d\u00f4le\u017eit\u00e9 predpoklady, ktor\u00e9 m\u00f4\u017eu urobi\u0165 tento pr\u00edstup k organiz\u00e1cii viac alebo menej spo\u013eahliv\u00fdm v niektor\u00fdch spolo\u010dnostiach. Proces iter\u00e1cie zni\u017euje n\u00e1klady na pokusy a omyly. Ak s\u00fa n\u00e1klady na chybu vysok\u00e9, mo\u017eno budete chcie\u0165 ich minimalizova\u0165 (ale to sa neodpor\u00fa\u010da pre medic\u00ednske aplik\u00e1cie alebo v\u00fdrobu). Okrem toho, ak pracujete s petabajtmi alebo exabajtmi \u00fadajov, m\u00f4\u017ee by\u0165 potrebn\u00e1 \u0161pecializ\u00e1cia v oblasti n\u00e1vrhu \u00fadajov. Rovnako, ak je udr\u017eanie obchodn\u00fdch pr\u00edle\u017eitost\u00ed online a ich dostupnos\u0165 d\u00f4le\u017eitej\u0161ia ako ich zdokona\u013eovanie, funk\u010dn\u00e9 prek\u00e1\u017eky m\u00f4\u017eu prev\u00e1\u017ei\u0165 nad u\u010den\u00edm. Nakoniec model pln\u00e9ho z\u00e1sobn\u00edka sa opiera o n\u00e1zory odborn\u00edkov. Nie s\u00fa to jednoro\u017ece; daj\u00fa sa n\u00e1js\u0165 alebo pripravi\u0165 samostatne. Av\u0161ak, s\u00fa ve\u013emi \u017eiadan\u00ed, a na pril\u00e1kanie a udr\u017eanie t\u00fdchto odborn\u00edkov vo firm\u00e1ch bude potrebn\u00e1 konkuren\u010dn\u00e1 materi\u00e1lna kompenz\u00e1cia, udr\u017eate\u013en\u00e9 firemn\u00e9 hodnoty a zauj\u00edmav\u00e1 pr\u00e1ca. Uistite sa, \u017ee va\u0161a firemn\u00e1 kult\u00fara dok\u00e1\u017ee zabezpe\u010di\u0165 tak\u00e9to podmienky.<\/p>\n

Aj s oh\u013eadom na v\u0161etko povedan\u00e9, domnievam sa, \u017ee model pln\u00e9ho z\u00e1sobn\u00edka poskytuje najlep\u0161ie podmienky na za\u010datie. Za\u010dnite s nimi a potom vedome prejdite k funk\u010dn\u00e9mu rozdeleniu pr\u00e1ce len vtedy, ke\u010f je to naozaj potrebn\u00e9.<\/p>\n

Existuj\u00fa aj \u010fal\u0161ie nev\u00fdhody funk\u010dnej \u0161pecializ\u00e1cie. M\u00f4\u017ee to vies\u0165 k strate zodpovednosti a pasivite zamestnancov. Sam Smith kritizuje rozdelenie pr\u00e1ce s t\u00fdm, \u017ee vedie k otupeniu talentu, t. j. zamestnanci sa st\u00e1vaj\u00fa neznal\u00fdmi a uzavret\u00fdmi, preto\u017ee ich r\u00f4zne \u00falohy s\u00fa obmedzen\u00e9 na nieko\u013eko opakuj\u00facich sa \u010dinnost\u00ed. Zatia\u013e \u010do \u0161pecializ\u00e1cia m\u00f4\u017ee zabezpe\u010di\u0165 efektivitu procesu, zriedkakedy in\u0161piruje zamestnancov.<\/p>\n

Univerz\u00e1lne \u00falohy zabezpe\u010duj\u00fa v\u0161etko, \u010do podnecuje spokojnos\u0165 s pr\u00e1cou: autonomia, odborn\u00e9 zru\u010dnosti a zameranie na cie\u013e. Autonomia spo\u010d\u00edva v tom, \u017ee nie s\u00fa z\u00e1visl\u00e9 od nie\u010doho in\u00e9ho pri dosahovan\u00ed \u00faspechu. Odborn\u00e9 zru\u010dnosti sa t\u00fdkaj\u00fa siln\u00fdch konkuren\u010dn\u00fdch v\u00fdhod. A zameranie na cie\u013e - v mo\u017enosti ovplyvni\u0165 podnikanie, ktor\u00e9 vytv\u00e1raj\u00fa. Ak sa n\u00e1m podar\u00ed zauja\u0165 \u013eud\u00ed ich pr\u00e1cou a ovplyvni\u0165 firmu, v\u0161etko ostatn\u00e9 sa dostane na svoje miesto.
\n
Zdroj: habr.com<\/a><\/p>","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"excerpt":{"rendered":"

HIROSHI WATANABE\/GETTY IMAGES \u0412 \u043a\u043d\u0438\u0433\u0435 \u00ab\u0411\u043e\u0433\u0430\u0442\u0441\u0442\u0432\u043e \u043d\u0430\u0440\u043e\u0434\u043e\u0432\u00bb \u0410\u0434\u0430\u043c \u0421\u043c\u0438\u0442 \u043f\u043e\u043a\u0430\u0437\u044b\u0432\u0430\u0435\u0442, \u043a\u0430\u043a \u0440\u0430\u0437\u0434\u0435\u043b\u0435\u043d\u0438\u0435 \u0442\u0440\u0443\u0434\u0430 \u0441\u0442\u0430\u043d\u043e\u0432\u0438\u0442\u0441\u044f \u0433\u043b\u0430\u0432\u043d\u044b\u043c \u0438\u0441\u0442\u043e\u0447\u043d\u0438\u043a\u043e\u043c \u043f\u043e\u0432\u044b\u0448\u0435\u043d\u0438\u044f \u043f\u0440\u043e\u0438\u0437\u0432\u043e\u0434\u0438\u0442\u0435\u043b\u044c\u043d\u043e\u0441\u0442\u0438. \u041f\u0440\u0438\u043c\u0435\u0440\u043e\u043c \u0432\u044b\u0441\u0442\u0443\u043f\u0430\u0435\u0442 \u043b\u0438\u043d\u0438\u044f \u0441\u0431\u043e\u0440\u043a\u0438 \u0437\u0430\u0432\u043e\u0434\u0430 \u043f\u043e \u043f\u0440\u043e\u0438\u0437\u0432\u043e\u0434\u0441\u0442\u0432\u0443 \u0431\u0443\u043b\u0430\u0432\u043e\u043a: \u00ab\u041e\u0434\u0438\u043d \u0440\u0430\u0431\u043e\u0447\u0438\u0439 \u0442\u044f\u043d\u0435\u0442 \u043f\u0440\u043e\u0432\u043e\u043b\u043e\u043a\u0443, \u0434\u0440\u0443\u0433\u043e\u0439 \u0432\u044b\u043f\u0440\u044f\u043c\u043b\u044f\u0435\u0442 \u0435\u0451, \u0442\u0440\u0435\u0442\u0438\u0439 \u043e\u0431\u0440\u0435\u0437\u0430\u0435\u0442, \u0447\u0435\u0442\u0432\u0435\u0440\u0442\u044b\u0439 \u0437\u0430\u043e\u0441\u0442\u0440\u044f\u0435\u0442 \u043a\u043e\u043d\u0435\u0446, \u043f\u044f\u0442\u044b\u0439 \u043e\u0431\u0442\u0430\u0447\u0438\u0432\u0430\u0435\u0442 \u0434\u0440\u0443\u0433\u043e\u0439 \u043a\u043e\u043d\u0435\u0446 \u0434\u043b\u044f \u043d\u0430\u0441\u0430\u0436\u0438\u0432\u0430\u043d\u0438\u044f \u0433\u043e\u043b\u043e\u0432\u043a\u0438\u00bb. \u0411\u043b\u0430\u0433\u043e\u0434\u0430\u0440\u044f \u0441\u043f\u0435\u0446\u0438\u0430\u043b\u0438\u0437\u0430\u0446\u0438\u0438, \u043e\u0440\u0438\u0435\u043d\u0442\u0438\u0440\u043e\u0432\u0430\u043d\u043d\u043e\u0439 \u043d\u0430 \u043e\u043f\u0440\u0435\u0434\u0435\u043b\u0435\u043d\u043d\u044b\u0435 \u0444\u0443\u043d\u043a\u0446\u0438\u0438, \u043a\u0430\u0436\u0434\u044b\u0439 \u0440\u0430\u0431\u043e\u0442\u043d\u0438\u043a \u0441\u0442\u0430\u043d\u043e\u0432\u0438\u0442\u0441\u044f \u0432\u044b\u0441\u043e\u043a\u043e\u043a\u0432\u0430\u043b\u0438\u0444\u0438\u0446\u0438\u0440\u043e\u0432\u0430\u043d\u043d\u044b\u043c […]<\/p>\n","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"author":1,"featured_media":24671,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[702],"tags":[],"class_list":["post-32907","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news"],"aioseo_notices":[],"aioseo_head":"\n\t\t\n\t\n\t\n\t\n\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\n","aioseo_head_json":{"title":"\ud83e\udd47Pre\u010do potrebuj\u00fa t\u00edmy Data Science univerz\u00e1lov, a nie \u0161pecialistov | ProHoster","description":"HIROSHI WATANABE\/GETTY.","canonical_url":"https:\/\/prohoster.info\/sk\/blog\/news\/pochemu-data-science-komandam-nuzhny-universaly-a-ne-spetsialisty","robots":"max-image-preview:large","keywords":"","webmasterTools":{"miscellaneous":""},"schema":null,"og:locale":"sk_SK","og:site_name":"ProHoster | \u041a\u0443\u043f\u0438\u0442\u044c \u043d\u0430\u0434\u0435\u0436\u043d\u044b\u0439 \u0445\u043e\u0441\u0442\u0438\u043d\u0433 \u0434\u043b\u044f \u0441\u0430\u0439\u0442\u043e\u0432 \u0441 \u0437\u0430\u0449\u0438\u0442\u043e\u0439 \u043e\u0442 DDoS, VPS VDS \u0441\u0435\u0440\u0432\u0435\u0440\u044b","og:type":"article","og:title":"\ud83e\udd47\u041f\u043e\u0447\u0435\u043c\u0443 Data Science \u043a\u043e\u043c\u0430\u043d\u0434\u0430\u043c \u043d\u0443\u0436\u043d\u044b \u0443\u043d\u0438\u0432\u0435\u0440\u0441\u0430\u043b\u044b, \u0430 \u043d\u0435 \u0441\u043f\u0435\u0446\u0438\u0430\u043b\u0438\u0441\u0442\u044b | ProHoster","og:description":"HIROSHI WATANABE\/GETTY.","og:url":"https:\/\/prohoster.info\/sk\/blog\/news\/pochemu-data-science-komandam-nuzhny-universaly-a-ne-spetsialisty","og:image":"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg","og:image:secure_url":"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg","og:image:width":350,"og:image:height":350,"article:published_time":"2019-10-31T18:49:39+00:00","article:modified_time":"2019-10-31T18:49:39+00:00","article:publisher":"https:\/\/www.facebook.com\/prohoster","article:author":"https:\/\/www.facebook.com\/prohoster"},"aioseo_meta_data":{"post_id":"32907","title":null,"description":null,"keywords":null,"keyphrases":null,"primary_term":null,"canonical_url":null,"og_title":null,"og_description":null,"og_object_type":"default","og_image_type":"default","og_image_url":null,"og_image_width":null,"og_image_height":null,"og_image_custom_url":null,"og_image_custom_fields":null,"og_video":null,"og_custom_url":null,"og_article_section":null,"og_article_tags":null,"twitter_use_og":false,"twitter_card":"default","twitter_image_type":"default","twitter_image_url":null,"twitter_image_custom_url":null,"twitter_image_custom_fields":null,"twitter_title":null,"twitter_description":null,"schema":{"blockGraphs":[],"customGraphs":[],"default":{"data":{"Article":[],"Course":[],"Dataset":[],"FAQPage":[],"Movie":[],"Person":[],"Product":[],"ProductReview":[],"Car":[],"Recipe":[],"Service":[],"SoftwareApplication":[],"WebPage":[]},"graphName":"","isEnabled":true},"graphs":[]},"schema_type":null,"schema_type_options":null,"pillar_content":false,"robots_default":true,"robots_noindex":false,"robots_noarchive":false,"robots_nosnippet":false,"robots_nofollow":false,"robots_noimageindex":false,"robots_noodp":false,"robots_notranslate":false,"robots_max_snippet":null,"robots_max_videopreview":null,"robots_max_imagepreview":"large","priority":null,"frequency":null,"local_seo":null,"seo_analyzer_scan_date":"2026-01-21 13:06:20","breadcrumb_settings":null,"limit_modified_date":false,"reviewed_by":null,"ai":null,"created":"2021-03-01 02:50:24","updated":"2026-01-21 13:06:20","focus_keyword":null,"additional_keywords":null,"truseo_locale":null},"gt_translate_keys":[{"key":"link","format":"url"}],"_links":{"self":[{"href":"https:\/\/prohoster.info\/sk\/wp-json\/wp\/v2\/posts\/32907","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/prohoster.info\/sk\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/prohoster.info\/sk\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/prohoster.info\/sk\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/prohoster.info\/sk\/wp-json\/wp\/v2\/comments?post=32907"}],"version-history":[{"count":0,"href":"https:\/\/prohoster.info\/sk\/wp-json\/wp\/v2\/posts\/32907\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/prohoster.info\/sk\/wp-json\/wp\/v2\/media\/24671"}],"wp:attachment":[{"href":"https:\/\/prohoster.info\/sk\/wp-json\/wp\/v2\/media?parent=32907"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/prohoster.info\/sk\/wp-json\/wp\/v2\/categories?post=32907"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/prohoster.info\/sk\/wp-json\/wp\/v2\/tags?post=32907"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}