{"id":34123,"date":"2019-10-31T21:56:30","date_gmt":"2019-10-31T18:56:30","guid":{"rendered":"https:\/\/prohoster.info\/blog\/qa-hakatony\/"},"modified":"2019-10-31T21:56:30","modified_gmt":"2019-10-31T18:56:30","slug":"qa-hakatony","status":"publish","type":"post","link":"https:\/\/prohoster.info\/sq\/blog\/news\/qa-hakatony","title":{"rendered":"QA: Hackathon","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"

\"QA:
\n
\nPjesa p\u00ebrfundimtare e trilogjis\u00eb p\u00ebr hackathonet. N\u00eb pjes\u00ebn e par\u00eb<\/a><\/noindex> tregova p\u00ebr motivimin p\u00ebr t\u00eb marr\u00eb pjes\u00eb n\u00eb k\u00ebto ngjarje. Pjesa e dyt\u00eb<\/a><\/noindex> ishte e p\u00ebrkushtuar p\u00ebr gabimet e organizator\u00ebve dhe pasojat e tyre. Pjesa p\u00ebrfundimtare do t\u00eb p\u00ebrgjigjet n\u00eb pyetje q\u00eb nuk u p\u00ebrfshin\u00eb n\u00eb dy pjes\u00ebt e para. <\/a><\/noindex><\/p>\n

T\u00eb tregoj si fillove t\u00eb marr\u00ebsh pjes\u00eb n\u00eb hackathonet. <\/b>
\nIsha duke studiuar n\u00eb masterin e Universitetit t\u00eb Lappeenranta nd\u00ebrkoh\u00eb q\u00eb merrja pjes\u00eb n\u00eb garat p\u00ebr analiz\u00ebn e t\u00eb dh\u00ebnave. Dita ime tipike dukej k\u00ebshtu: zgjohem n\u00eb or\u00ebn 8, disa or\u00eb n\u00eb universitet, m\u00eb pas gara dhe kurse deri n\u00eb mesnat\u00eb (nd\u00ebrsa prisja dor\u00ebzimin, shikoja ligj\u00ebrata ose lexoja artikuj). Ky grafik i ngjeshur dha fruta, dhe fitova gar\u00ebn p\u00ebr analiz\u00ebn e t\u00eb dh\u00ebnave MERC-2017 (p\u00ebr k\u00ebt\u00eb madje kishim nj\u00eb postim n\u00eb Habr\u00e9<\/a><\/noindex>). Fitorja m\u00eb dha besim n\u00eb vetvete, dhe kur rast\u00ebsisht pata informacion p\u00ebr hackathonin SkinHack 2 n\u00eb Mosk\u00eb, vendosa t\u00eb vizitoj prind\u00ebrit dhe p\u00ebr nj\u00ebher\u00eb t\u00eb zbuloj se \u00e7far\u00eb \u00ebsht\u00eb hackathoni. <\/p>\n

Hackathoni doli t\u00eb ishte mjaft arg\u00ebtues. Kishte dy rrjedha p\u00ebr analiz\u00ebn e t\u00eb dh\u00ebnave me metrika t\u00eb qarta dhe nj\u00eb dataset me \u00e7mime prej 100k rubla. Rrjedha e tret\u00eb ishte p\u00ebr zhvillimin e aplikacioneve me \u00e7mim prej 50k, dhe n\u00eb t\u00eb nuk kishte pjes\u00ebmarr\u00ebs. N\u00eb nj\u00eb moment, organizatori tha se nj\u00eb dritare me nj\u00eb buton pa funksionalitet mund t\u00eb fitoj\u00eb 50k, sepse \u00e7mimi nuk mund t\u00eb mos jepet. Nuk vendosa t\u00eb m\u00ebsoj t\u00eb programoj aplikacione (nuk garoj atje ku mund t\u00eb <\/p>\n

kalohem leht\u00eb), por p\u00ebr mua ishte nj\u00eb mesazh i qart\u00eb se fusha n\u00eb hackathonet nuk \u00ebsht\u00eb e mbushur. At\u00ebher\u00eb un\u00eb isha duke zgjidhur t\u00eb dy rrjedhat p\u00ebr analiz\u00ebn e t\u00eb dh\u00ebnave vet\u00eb. Gjeta nj\u00eb lik n\u00eb t\u00eb dh\u00ebna, i cili lejonte t\u00eb arrinim nj\u00eb pik\u00eb ideale, por kolona me lik nuk ishte n\u00eb t\u00eb dh\u00ebnat testuese q\u00eb mora dy or\u00eb para p\u00ebrfundimit t\u00eb ngjarjes (p\u00ebr ta th\u00ebn\u00eb, at\u00ebher\u00eb kuptova se prania e kolon\u00ebs \"target\" n\u00eb trajnim nuk merret parasysh si lik). N\u00eb t\u00eb nj\u00ebjt\u00ebn koh\u00eb, u hap leaderboardi, dor\u00ebzimi im pa lik zinte vendin e tret\u00eb nga pes\u00eb, kishte nj\u00eb diferenc\u00eb t\u00eb madhe deri n\u00eb vendin e par\u00eb dhe vendosa t\u00eb mos humb koh\u00eb dhe u largova. <\/p>\n

Pas pas se kam analizova m\u00eb n\u00eb qet\u00ebsi se \u00e7far\u00eb ndodhi, gjet\u00ebm shum\u00eb gabime (nj\u00eb nga zakonet e mia \u00ebsht\u00eb t\u00eb rishikoj n\u00eb mendje se \u00e7far\u00eb ndodhi me nj\u00eb blloknot dhe t\u00eb analizoj gabimet, shkakun e tyre dhe \u00e7far\u00eb mund t\u00eb ishte ndryshuar - nj\u00eb trash\u00ebgimi e k\u00ebndshme e loj\u00ebs gjysm\u00eb-profesionale t\u00eb pokerit). Por nj\u00eb gj\u00eb ishte e qart\u00eb - ka shum\u00eb vler\u00eb n\u00eb hackathon-e, dhe un\u00eb thjesht duhet ta realizoj at\u00eb. Pas k\u00ebtij ngjarjeje, fillova t\u00eb monitoroj aktivitete dhe grupe, dhe hackathon-i pasues nuk ishte i gjat\u00eb p\u00ebr t'u pritur. Pastaj nj\u00eb tjet\u00ebr, dhe nj\u00eb tjet\u00ebr... <\/p>\n

Pse merresh me hackathon-e dhe jo me Kaggle?<\/b>
\nAktualisht, nuk m\u00eb p\u00eblqen Kaggle. Nga nj\u00eb nivel t\u00eb caktuar k\u00ebndi, pa arsye specifike pjes\u00ebmarrjeje, Kaggle b\u00ebhet m\u00eb pak i dobish\u00ebm se aktivitete t\u00eb tjera. Kam marr\u00eb pjes\u00eb shum\u00eb m\u00eb par\u00eb, duket se kam arritur n\u00eb nj\u00ebfar\u00eb m\u00ebnyre t\u00eb \u201cshk\u00ebputem\u201d. <\/p>\n

Pse hackathon-e dhe jo pun\u00eb n\u00eb projektin t\u00ebnd? <\/b>
\nM\u00eb p\u00eblqen ideja p\u00ebr t\u00eb b\u00ebr\u00eb di\u00e7ka t\u00eb shk\u00eblqyer me duar t\u00eb mia n\u00eb nj\u00eb rit\u00ebm jo t\u00eb shpejt\u00eb. Djemt\u00eb nga ODS organizuan ODS pet projects<\/a><\/noindex> p\u00ebr t\u00eb gjith\u00eb ata q\u00eb duan t\u00eb punojn\u00eb n\u00eb projektin e tyre gjat\u00eb fundjav\u00ebs n\u00eb nj\u00eb grup shok\u00ebsh. Mendoj se shum\u00eb shpejt do t'u bashkohem atyre. <\/p>\n

Si i gjen aktivitetet? <\/b>
\nBurimi kryesor \u00ebsht\u00eb hackathon.com (bot\u00eb) dhe grupi n\u00eb Telegram Russian Hackers<\/a><\/noindex> (Rusi). P\u00ebrve\u00e7 k\u00ebsaj, njoftimet p\u00ebr aktivitete shpesh shfaqen n\u00eb reklama n\u00eb rrjetet sociale dhe n\u00eb LinkedIn. N\u00ebse nuk gjen asgj\u00eb, mund t\u00eb shihni k\u00ebtu: mlh.io, devpost.com, hackevents.co, hackalist.org, HackathonsNear.me, hackathon.io. <\/p>\n

A e p\u00ebrgatit nj\u00eb plan zgjidhjeje para pjes\u00ebmarrjes apo gjith\u00e7ka vendoset n\u00eb moment? P\u00ebr shembull, nj\u00eb jav\u00eb para hackathon-it mendon: \"K\u00ebtu do t\u00eb nevojitet nj\u00eb specialist i till\u00eb dhe i till\u00eb, do t\u00eb duhet t\u00eb k\u00ebrkojm\u00eb\"?<\/b>
\nN\u00ebse hackathon-i \u00ebsht\u00eb produktiv - po, p\u00ebrgatitem. Disa jav\u00eb para, mend t'i jap di\u00e7ka se \u00e7far\u00eb do t\u00eb b\u00ebj, mendoj kush mund t\u00eb jet\u00eb i dobish\u00ebm, mbledh nj\u00eb ekip nga miqt\u00eb ose pjes\u00ebmarr\u00ebsit e hackathon-eve t\u00eb kaluara. <\/p>\n

A \u00ebsht\u00eb e mundur t\u00eb hackosh hackathon-in vet\u00ebm? \u00c7far\u00eb t\u00eb b\u00ebsh n\u00ebse nuk ke ekip? <\/b>
\nHackathonet e data science jan\u00eb realisht (un\u00eb jam nj\u00eb shembull i gjall\u00eb), por s'kam par\u00eb hackathon p\u00ebr produkte, megjithat\u00eb mendoj se po. Fatkeq\u00ebsisht, ndonj\u00ebher\u00eb organizator\u00ebt vendosin kufizime p\u00ebr numrin minimal t\u00eb pjes\u00ebmarr\u00ebsve n\u00eb ekip. Mendoj se kjo ndodh sepse jo t\u00eb gjith\u00eb \"vetmitar\u00ebt\" arrijn\u00eb n\u00eb finale (dometh\u00ebn\u00eb, thjesht heqin dor\u00eb p\u00ebrball\u00eb v\u00ebshtir\u00ebsive t\u00eb para), pjes\u00ebmarrja n\u00eb ekip ndihmon gjithsesi. Pas evenimentit, pritet q\u00eb t\u00eb vazhdoni pun\u00ebn mbi projektin. Me ekipin, do t\u00eb jet\u00eb m\u00eb e leht\u00eb ta \u00e7oni projektin deri n\u00eb fund. <\/p>\n

N\u00eb p\u00ebrgjith\u00ebsi, un\u00eb rekomandoj gjithmon\u00eb t\u00eb merrni pjes\u00eb me ekip. N\u00ebse nuk keni ekip, organizator\u00ebt gjithmon\u00eb ndihmojn\u00eb p\u00ebr ta gjetur ose krijuar nj\u00eb.<\/p>\n

Si arrini t\u00eb p\u00ebrballoni lodhjen gjat\u00eb hackathonit? <\/b>
\nN\u00eb hackathon jepet 2 dit\u00eb p\u00ebr pun\u00eb, q\u00eb jan\u00eb 48 or\u00eb (30-48 or\u00eb, do t\u00eb marrim 48 p\u00ebr thjesht\u00ebsi llogarit\u00ebse). Heqim koh\u00ebn p\u00ebr gjum\u00eb (16-20 or\u00eb), mbeten jo m\u00eb shum\u00eb se 30. Nga k\u00ebto, realisht do t\u00eb shpenzohen rreth 8 or\u00eb p\u00ebr pun\u00eb produktive (n\u00eb mesatar). N\u00ebse e organizoni mir\u00eb pun\u00ebn (gjumi, ushqimi, dalja n\u00eb aj\u00ebr t\u00eb past\u00ebr, aktivitetet fizike, momentet e vet\u00ebdijes, komunikimi i duhur me ekipin dhe ndryshimi i aktiviteteve), at\u00ebher\u00eb or\u00ebt e thell\u00eb t\u00eb pun\u00ebs mund t\u00eb arrijn\u00eb deri n\u00eb 12-14. Pas k\u00ebsaj pune do t\u00eb ndiheni t\u00eb shteruar, por do t\u00eb jet\u00eb nj\u00eb lodhje e k\u00ebndshme. T\u00eb kodosh pa gjum\u00eb dhe pa pushime duke u mb\u00ebshtetur n\u00eb energjizues \u00ebsht\u00eb nj\u00eb rrug\u00eb drejt d\u00ebshtimit.<\/p>\n

A keni pipeline t\u00eb gatshme p\u00ebr hackathonet? Si jan\u00eb shfaqur ato, si jan\u00eb organizuar (skedar\u00ebt .py jan\u00eb t\u00eb ndar\u00eb n\u00eb dosje, secili p\u00ebr detyr\u00ebn e tij, etj.) dhe si t\u00eb filloni t\u00eb krijoni vet\u00eb t\u00eb tilla?<\/b>
\nNuk p\u00ebrdor zgjidhje t\u00eb plota nga hackathonet e kaluara n\u00eb t\u00eb reja, por kam t\u00eb vetin nj\u00eb zoo modelet dhe pipeline nga garat e kaluara. Nuk kam nevoj\u00eb t\u00eb rishkruaj pjes\u00eb standarde nga e para (p\u00ebr shembull, encoding t\u00eb sakt\u00eb t\u00eb targetit ose nj\u00eb rrjet t\u00eb thjesht\u00eb p\u00ebr identifikimin e q\u00ebllimit nga teksti), \u00e7ka m\u00eb kursen shum\u00eb koh\u00eb. <\/p>\n

N\u00eb k\u00ebt\u00eb moment, kjo duket k\u00ebshtu: p\u00ebr \u00e7do gar\u00eb apo hackathon ka nj\u00eb repo t\u00eb vet n\u00eb GitHub, ku ruhet notebook-\u00ebt, skenaret dhe nj\u00eb dokumentacion i vog\u00ebl mbi at\u00eb q\u00eb ndodh. Plus, ka nj\u00eb repo t\u00eb ve\u00e7ant\u00eb p\u00ebr \"truket\" e ndryshme (si p\u00ebr shembull encoding t\u00eb sakt\u00eb t\u00eb targetit me cross-validation). Nuk mendoj se \u00ebsht\u00eb zgjidhja m\u00eb elegante, por p\u00ebr momentin m\u00eb k\u00ebnaq. <\/p>\n

Do t\u00eb filloja me ruajtjen e t\u00eb gjith\u00eb kodit tim n\u00eb luks dhe shkruajn\u00eb nj\u00eb dokumentacion t\u00eb shkurt\u00ebr (pse, \u00e7far\u00eb, si kam b\u00ebr\u00eb dhe rezultati). <\/p>\n

A \u00ebsht\u00eb e mundur t\u00eb p\u00ebrgatisim MVP nga e para n\u00eb nj\u00eb koh\u00eb kaq t\u00eb shkurt\u00ebr, apo t\u00eb gjitha pjes\u00ebmarr\u00ebsit vijn\u00eb me zgjidhje t\u00eb gatshme?<\/b>
\nMund t\u00eb flas vet\u00ebm p\u00ebr projektet q\u00eb lidhen me datascience \u2014 po, \u00ebsht\u00eb e mundur. P\u00ebr mua, MVP \u00ebsht\u00eb nj\u00eb p\u00ebrb\u00ebrje e dy faktor\u00ebve: <\/p>\n