Kedro chirongwa GitHub peji
Kedro ndiyo yekutanga yakavhurwa sosi chishandiso chakagadziriswa nechikamu chemakambani ekubvunza McKinsey. Yakagadzirwa kuti ishandiswe nevesainzi data uye mainjiniya. Ndiyo kodhi raibhurari iyo inogona kushandiswa kugadzira data nemapombi, zvivakwa zvekuvaka muchina chirongwa chekudzidza.
McKinsey & Kambani ikambani yeAmerican yepasi rese yekutarisa kambani. Inoita yepamusoro uye huwandu hwekuongorora kuti uongorore manejimendi sarudzo mumasangano eruzhinji neakazvimirira. Vatengi vake vanosanganisira 80% yemakambani makuru epasirese.
Kutanga yakavhurwa sosi chishandiso
Iyo kambani yakanga isati yamboburitsa imwe yemidziyo yakagadzirwa mukati-imba pasi peyakavhurika sosi rezinesi. Muchokwadi, Kedro akaberekwa seyemidziyo software. Nekudaro, kana hukama nekambani hwapera, vatengi vaive vasisakwanise kuwana chirongwa ichi.
Zita rekuti Kedro rinobva ku izwi rechiGiriki rekuti centre kana musimboti. Yakasarudzwa nekuti ichi chakavhurika sosi yekushandisa inopa yakakosha kodhi yekuburitsa epamberi ongororo mapurojekiti.
Kedro ine mikana miviri mikuru:
- Inoita kuti zvikwata zvibatane zviri nyore nekugadzirisa kodhi yekuongorora nenzira yunifomu.
- Iyo inobvumira zvese zvikamu kuti zviyerere zvisina mushe kuburikidza nematanho ese eprojekiti.
Izvi zvinosanganisira
- Kubatanidzwa kwemasosi ezvinyorwa,
- Kuchenesa data
- Ratidza Kusika
- Fudza iyo data mumakina ekudzidza mamodheru ekutsanangudza kana kufanofungidzira ongororo.
Kedro zvakare inobatsira kuburitsa yakagadzirira-kushandisa-kodhi. Izvi zvinoita kuti ive inoshanda chaizvo kune data masayendisiti asiri anowanzo kuve nyanzvi mukugadzirwa kwesoftware.
Nei Kedro ichibatsira?
Vhura zvishandiso senge Kedro tendera kuderedza nguva inotora kushandura prototype kuita kodhi yekugadzira nemavhiki. Vaongorori vanogona kushandisa shoma nguva kukodha uye yakawanda nguva kugadzirisa vatengi vavo.
Kedro inobatsira zvikwata kugadzira modular data chiteshi, kuyedzwa, kuberekwazve mune chero nharaunda uye yakashandurwa, ichibvumira vashandisi kuwana yapfuura data state. Iyo imwecheteyo kodhi inogona kuenda kubva kune imwechete yekuvandudza laptop kune ye bhizinesi-giredhi chirongwa uchishandisa gore komputa. Iyo inogona zvakare kushandiswa neese maindasitiri, mamodheru uye dhata zvinyorwa.
McKinsey akatoshandisa Kedro pane anopfuura makumi mashanu mapurojekiti kusvika parizvino. Sekureva kwemumwe mukuru, vatengi kunyanya senge kuona kwemaipi. Ivo vanobva vaona iwo akasiyana masiteji eshanduko, mhando dzemamodeli dzakabatanidzwa, uye dzinogona kuronda mhedzisiro kumashure kune yakasvibira dhata sosi.
McKinsey kwete yekutanga kambani isina hukama chaihwo netekinoroji iyo inoburitsa yakavhurika sosi maturusi. Uber neAirbnb vanga vatozviita.
Kendro Zvimiro uye Kuisirwa
Kedro chishandiso chekufambisa chekufambisa che kugadzirwa kwemasimba akasimba, anotyisa, anotumirwa, anoberekwazve uye akagadziridzwa.
Ndeapi maitiro makuru eKedro?
1. Project template uye kukodha zviyero
- Iyo yakapusa-kushandisa, yakajairwa chirongwa template
- Zvirongwa zvehunyanzvi, kunyoreswa, kurodha dhata uye Jupyter Notebooks / Lab.
- Bvunzo-inotungamirwa kusimudzira uchishandisa pytest
- Sphinx kusangana kuti ibudise yakanyorwa zvakanaka kodhi
2. Kuburitsa dhata uye kushandura
- Kuparadzaniswa kweiyo komputa dura kubva kune dhata manejimendi manejimendi, kusanganisira rutsigiro rwemhando dzakasiyana dzedata uye sarudzo dzekuchengetedza
- Mavhezheni eiyo yako data seti uye muchina wekufunda mamodheru
3. Modularity uye kubvisa mapaipi
- Tsigiro yezvakachena Python mabasa, node, kupatsanura hombe machunks ekodhi muzvikamu zvidiki zvakazvimirira.
- Otomatiki resolution yekutsamira pakati penzvimbo
4. Kuwedzera kwezvinhu
- Iyo plugin system inobaya mirairo muKedro's command line interface (CLI): Kedro-Mhepo inoyerera, zvichiita kuti zvive nyore kuratidzira yako data pombi muKedro usati waiendesa kuAirflow, iyo yekushambadzira kuronga. Kedro-Docker, chishandiso kurongedza uye kutumira Kedro mapurojekiti mumidziyo
- Kedro inogona kuendeswa munharaunda, panzvimbo uye mugore (AWS, Azure, uye GCP) kana mumasumbu (EMR, Azure HDinsight, GCP, uye Databricks).
Tinogona kuisa Kedro pane yedu yekutanga-yakatarwa kugoverwa kweLinux nekuita:
sudo apt install python3-pip
pip install kedro
Kugadzirisa:
pip3 install kedro -U
Tinogona kuona zvinyorwa ne:
kedro docs
Mamwe mashoko anogona kuwanikwa pa peji rerojekiti