IKedro project GitHub ikhasi
IKedro iyithuluzi lokuqala lomthombo ovulekile elenziwe ukuhlukaniswa kwenkampani yabeluleki iMcKinsey. Yadalelwa ukuthi isetshenziswe ososayensi nedatha. Ingabe i- library library engasetshenziswa ukudala idatha namapayipi, izakhiwo zokwakha iphrojekthi yokufunda ngomshini.
IMcKinsey & Inkampani iyinkampani yokubonisana nokuphathwa komhlaba wonke yaseMelika. Yenza ukuhlaziywa kwekhwalithi nobungako ukuhlola izinqumo zokuphatha emikhakheni kahulumeni neyangasese. Amakhasimende akhe afaka ama-80% ezinkampani ezinkulu emhlabeni.
Ithuluzi lokuqala lomthombo ovulekile
Le nkampani ayikaze ikhiphe elinye lamathuluzi athuthukiswe ngaphakathi endlini ngaphansi kwelayisense yomthombo ovulekile. Eqinisweni, uKedro wazalwa njengesoftware ephathelene. Kodwa-ke, lapho ubudlelwane nenkampani buphela, amakhasimende ayengasenakho ukufinyelela kulolu hlelo.
Igama elithi Kedro lisuselwa ku- igama lesiGreki lesikhungo noma umongo. Kukhethwe ngoba leli thuluzi lomthombo ovulekile linikeza ikhodi ebalulekile yokukhiqiza amaphrojekthi wokuhlaziya osezingeni eliphakeme.
IKedro inezinzuzo ezimbili eziyinhloko:
- Inika amandla amaqembu ukuthi asebenzisane kalula ngokwakha ikhodi yokuhlaziya ngendlela efanayo.
- Ivumela zonke izingxenye ukugeleza ngaphandle komthungo kuzo zonke izigaba zephrojekthi.
Lokhu kufaka phakathi
- Ukuhlanganiswa kwemithombo yedatha,
- Ukuhlanzwa kwedatha
- Ukudalwa kwesici
- Faka idatha kumamodeli wokufunda womshini wokuhlaziya okuchazayo noma kokuqagela.
UKedro naye isiza ukuletha ikhodi elungele ukusetshenziswa. Lokhu kwenza kube wusizo olukhulu kososayensi bedatha abangavamile ukuba ochwepheshe ekwakhiweni kwesoftware.
Kungani iKedro ilusizo?
Amathuluzi womthombo ovulekile njengeKedro allow ukunciphisa isikhathi esithathayo ukuguqula uhlobo oluthile lube yikhodi yokukhiqiza ngamasonto. Abahlaziyi bangachitha isikhathi esincane sokubhala ikhodi nesikhathi esiningi ukuxazulula inkinga kumakhasimende abo.
I-Kedro isiza amaqembu ukuthi enze iziteshi ze-modular zedatha, ezivivinywe, ezizalwe kabusha kunoma iyiphi imvelo futhi zivuselelwe, zivumela abasebenzisi ukuthi bakwazi ukufinyelela kwizizinda zedatha edlule. Leyo khodi ingasuka kukhompuyutha ephathekayo eyodwa iye kuphrojekthi ebangeni lebhizinisi isebenzisa i-computing yamafu. Ingasetshenziswa nazo zonke izimboni, amamodeli nemithombo yedatha.
UMcKinsey usevele usebenzise uKedro kumaphrojekthi angaphezu kwama-50 kuze kube manje. Ngokusho kwesinye isikhulu, amakhasimende ikakhulukazi athanda ukubonwa kwamapayipi. Ngokushesha babona izigaba ezahlukahlukene zenguquko, izinhlobo zamamodeli ezithintekayo, futhi bangakwazi ukulandela imiphumela emuva emthonjeni wedatha eluhlaza.
McKinsey akuyona inkampani yokuqala engahlobene ngqo nobuchwepheshe eshicilela amathuluzi wemithombo evulekile. I-Uber ne-Airbnb basebevele bekwenzile.
Izici zeKendro nokufakwa
IKedro iyithuluzi lokuthuthukisa ukuhamba komsebenzi ukwakhiwa kwamashaneli wedatha aqinile, asabekayo, atholakalayo, aphindaphindiwe futhi ahlaziyiwe.
Yiziphi izici eziyinhloko zeKedro?
1. Isifanekiso sephrojekthi namazinga wokufaka amakhodi
- Isifanekiso sephrojekthi esisebenziseka kalula
- Amasethingi obufakazi, ukubhaliswa, ukulayishwa kwedatha kanye neJupyter Notebooks / Lab.
- Intuthuko eqhutshwa ukuhlolwa kusetshenziswa i-pytest
- Ukuhlanganiswa kweSphinx ukukhiqiza ikhodi ebhalwe kahle
2. Isizinda sedatha nokwenza inguqulo
- Ukwehlukaniswa kwesendlalelo sekhompyutha kusuka kusendlalelo sokuphathwa kwedatha, kufaka phakathi ukusekelwa kwamafomethi wedatha ahlukahlukene nezinketho zokugcina.
- Izinhlobo zamasethi wakho wedatha namamodeli wokufunda womshini
3. Ukuhleleka nokukhishwa kwamapayipi
- Ukusekelwa kwemisebenzi yePython emsulwa, ama-node, ukwehlukanisa iziqeshana ezinkulu zekhodi zibe izingxenye ezincane ezizimele.
- Ukulungiswa okuzenzakalelayo kokuncika phakathi kwama-node
4. Ukwanda kwezici
- Uhlelo lwe-plugin olufaka imiyalo ku-interface yomugqa womyalo we-Kedro (CLI): I-Kedro-Airflow, okwenza kube lula ukwenza ipayipi ledatha yakho eKedro ngaphambi kokuyithumela ku-Airflow, isihleli sokuhamba komsebenzi. IKedro-Docker, ithuluzi lokupakisha nokuthumela amaphrojekthi we-Kedro ezitsheni
- IKedro ingasatshalaliswa endaweni, endaweni nasefwini (i-AWS, i-Azure, ne-GCP) noma ngamaqoqo (i-EMR, i-Azure HDinsight, i-GCP, ne-Databricks).
Singafaka i-Kedro ekusatshalalisweni kwethu kwe-Linux okwenziwe ngaphambilini ngokwenza:
sudo apt install python3-pip
pip install kedro
Ukwenza ngokoqobo:
pip3 install kedro -U
Sibona imibhalo nge:
kedro docs
Imininingwane engaphezulu ingatholakala ku- ikhasi lephrojekthi