Isikhathi sokubika ku-Excel siyashabalala ngokushesha - inkambiso ebheke kumathuluzi afanelekile wokwethula nokuhlaziya ulwazi ibonakala kuzo zonke izindawo. Kade sixoxa ngaphakathi nokwenziwa kwedijithali kokubika isikhathi eside futhi sakhetha i-Tableau visualization kanye nesistimu yezibalo zokuzisiza. U-Alexander Bezugly, inhloko yomnyango wezixazululo zokuhlaziya kanye nokubika we-M.Video-Eldorado Group, ukhulume ngolwazi nemiphumela yokwakha ideshibhodi yokulwa.
Ngizosho zisuka nje ukuthi akukhona konke okwakuhleliwe okwafezeka, kodwa okuhlangenwe nakho kwakuthakazelisa, ngithemba ukuthi kuyoba usizo nakuwe. Futhi uma kukhona onemibono yokuthi kungenziwa kanjani kangcono, ngingabonga kakhulu ngezeluleko nemibono yenu.
Ngezansi kokusikiwe kumayelana nalokho esihlangabezane nakho nesikufundile ngakho.
Siqale kuphi?
I-M.Video-Eldorado inemodeli yedatha ethuthukiswe kahle: ulwazi oluhlelekile olunokujula kwesitoreji okudingekayo kanye nenani elikhulu lemibiko yefomu elingashintshi (bona imininingwane eyengeziwe ). Kusukela kulawa, abahlaziyi benza amathebula e-pivot noma izincwadi zezindaba ezifomethwe ku-Excel, noma izethulo ezinhle ze-PowerPoint zabasebenzisi bokugcina.
Cishe eminyakeni emibili edlule, esikhundleni semibiko yefomu elingaguquki, saqala ukudala imibiko yokuhlaziya ku-SAP Analysis (isengezo se-Excel, empeleni ithebula le-pivot phezu kwenjini ye-OLAP). Kodwa leli thuluzi alikwazanga ukuhlangabezana nezidingo zabo bonke abasebenzisi; iningi liqhubekile nokusebenzisa ulwazi olucutshungulwe futhi ngabahlaziyi.
Abasebenzisi bethu bokugcina bawela ezigabeni ezintathu:
Abaphathi abaphezulu. Icela ulwazi ngendlela yethulwe kahle neqondakalayo.
Ukuphatha okuphakathi, abasebenzisi abathuthukile. Unentshisekelo ekuhloleni idatha futhi uyakwazi ukwakha imibiko ngokuzimela uma amathuluzi etholakala. Baba ngabasebenzisi abakhulu bemibiko yokuhlaziya ku-SAP Analysis.
Abasebenzisi abaningi. Abanantshisekelo yokuhlaziya idatha ngokuzimela; basebenzisa imibiko enezinga elilinganiselwe lenkululeko, ngefomethi yezincwadi zezindaba namathebula e-pivot ku-Excel.
Umbono wethu bekuwukumboza izidingo zabo bonke abasebenzisi futhi sibanike ithuluzi elilodwa, elikahle. Sinqume ukuqala ngabaphathi abaphezulu. Babedinga amadeshibhodi asebenziseka kalula ukuze bahlaziye imiphumela yebhizinisi ebalulekile. Ngakho-ke, siqale nge-Tableau futhi saqala sakhetha izikhombisi-ndlela ezimbili: izinkomba zokuthengisa nezokuthengisa eziku-inthanethi ezinokujula okulinganiselwe nobubanzi bokuhlaziya, okungahlanganisa cishe u-80% wedatha ecelwe abaphathi abaphezulu.
Njengoba abasebenzisi bamadeshibhodi babengabaphathi abaphezulu, enye i-KPI eyengeziwe yomkhiqizo yavela - isivinini sokuphendula. Akekho ozolinda imizuzwana engu-20-30 ukuze idatha ibuyekezwe. Ukuzulazula bekumele kwenziwe phakathi kwamasekhondi angu-4-5, noma okungcono, kwenziwe ngokushesha. Futhi thina, maye, sehlulekile ukufeza lokhu.
Nakhu indlela ukwakheka kwedeshibhodi yethu eyinhloko kwakubukeka ngayo:
Umqondo oyinhloko uwukuhlanganisa abashayeli abayinhloko be-KPI, okwakukhona abangu-19 sebebonke, ngakwesobunxele futhi bethule amandla abo okuguquguquka nokuqhekeka ngezici eziyinhloko kwesokudla. Umsebenzi ubonakala ulula, ukubona ngeso lokhozi kunengqondo futhi kuyaqondakala, uze ungene emininingwaneni.
Imininingwane 1. Ivolumu yedatha
Ithebula lethu elikhulu lokuthengisa lonyaka lithatha imigqa engaba yizigidi ezingu-300. Njengoba kudingekile ukukhombisa amandla onyaka odlule kanye nonyaka odlule, umthamo wedatha ekuthengisweni kwangempela kuphela cishe imigqa eyizigidi eziyizinkulungwane ezingu-1. Ulwazi olumayelana nedatha ehleliwe kanye ne-block sales block nayo igcinwa ngokuhlukana. Ngakho-ke, nakuba sisebenzise i-columnar in-memory DB SAP HANA, isivinini sombuzo ngokukhethwa kwazo zonke izinkomba isonto elilodwa ukusuka kwisitoreji samanje endizeni cishe imizuzwana engu-15-20. Isixazululo sale nkinga siyaziphakamisa - ukwenziwa kwezinto ezibonakalayo okwengeziwe kwedatha. Kodwa futhi inezingibe, okwengeziwe ngazo ngezansi.
Imininingwane 2. Izinkomba ezingezona ezengeziwe
Ama-KPI ethu amaningi ahlanganiswe nenani lamarisidi. Futhi le nkomba imele COUNT DISTINCT yenani lemigqa (hlola izihloko) futhi ibonisa amanani ahlukene kuye ngezibaluli ezikhethiwe. Isibonelo, ukuthi le nkomba kanye nokuphuma kwayo kufanele kubalwe kanjani:
Ukuze wenze izibalo zakho zibe zilungile, ungakwazi:
- Bala izinkomba ezinjalo on the fly in the storage;
- Yenza izibalo kuwo wonke umthamo wedatha ku-Tableau, i.e. ngesicelo ku-Tableau, nikeza yonke idatha ngokuya ngezihlungi ezikhethiwe kugranularity yendawo yerisidi;
- Dala umbukiso owenziwe ngezinto ezibonakalayo lapho zonke izinkomba zizobalwa kuzo zonke izinketho zesampula ezinikeza imiphumela ehlukile engeyona engeziwe.
Kuyacaca ukuthi esibonelweni i-UTE1 ne-UTE2 yizibaluli ezibonakalayo ezimelela isigaba somkhiqizo. Lokhu akuyona into emile; ukuphathwa ngaphakathi kwenkampani kwenzeka ngakho, ngoba Abaphathi abahlukene banesibopho samaqembu emikhiqizo ehlukene. Sibe nezibuyekezo eziningi zomhlaba zalolu hlelo, lapho wonke amazinga ashintsha, lapho ubudlelwano bubuyekezwa, futhi amaphuzu ashintsha njalo, lapho iqembu elilodwa lisuka kwenye indawo liye kwenye. Ekubikeni okuvamile, konke lokhu kubalwa ngokundiza kusuka ezicini zezinto ezibonakalayo; esimweni sokwenziwa kwale datha, kuyadingeka ukwakha indlela yokulandelela izinguquko ezinjalo kanye nokulayisha kabusha idatha yomlando ngokuzenzakalelayo. Umsebenzi ongeyona into encane kakhulu.
Imininingwane 3. Ukuqhathaniswa kwedatha
Leli phuzu lifana nelangaphambili. Okubalulekile ukuthi lapho uhlaziya inkampani, kuyisiko ukwakha amazinga amaningana wokuqhathanisa nesikhathi esedlule:
Ukuqhathanisa nesikhathi esedlule (usuku nosuku, isonto nesonto, inyanga nenyanga)
Kulesi siqhathaniso, kucatshangwa ukuthi kuye ngenkathi ekhethwe umsebenzisi (isibonelo, isonto lama-33 lonyaka), kufanele sibonise amandla ngeviki lama-32; uma sikhethe idatha yenyanga, isibonelo, ngoMeyi. , khona-ke lokhu kuqhathanisa kuzobonisa amandla ngo-April.
Ukuqhathanisa nonyaka odlule
I-nuance eyinhloko lapha ukuthi uma uqhathanisa ngosuku nangesonto, awuthathi usuku olufanayo lwangonyaka odlule, i.e. awukwazi ukuvele ubeke unyaka wamanje ukhiphe owodwa. Kufanele ubheke usuku lwesonto oluqhathanisayo. Uma uqhathanisa izinyanga, ngokuphambene nalokho, udinga ukuthatha usuku lwekhalenda olufanayo lwangonyaka odlule. Kukhona futhi ama-nuances ane-leap years. Kumakhosombe okuqala, lonke ulwazi lusatshalaliswa emini; azikho izinkambu ezihlukene ezinamaviki, izinyanga, noma iminyaka. Ngakho-ke, ukuze uthole i-cross-section egcwele yokuhlaziya iphaneli, awudingi ukubala isikhathi esisodwa, isibonelo ngesonto, kodwa amaviki angu-4, bese uqhathanisa le datha, ubonise amandla, ukuphambuka. Ngokufanelekile, le logic yokwenza iziqhathaniso ku-dynamics ingaphinda isetshenziswe ku-Tableau noma ohlangothini lwesitolo sangaphambili. Yebo, futhi vele sazi futhi sacabanga ngale mininingwane esigabeni sokuklama, kodwa kwakunzima ukubikezela umthelela wabo ekusebenzeni kwedeshibhodi yokugcina.
Lapho sisebenzisa ideshibhodi, salandela indlela ende ye-Agile. Umsebenzi wethu bekuwukuhlinzeka ngethuluzi lokusebenza nedatha edingekayo yokuhlolwa ngokushesha okukhulu. Ngakho-ke, sahamba nge-sprints futhi saqala ekunciphiseni umsebenzi ohlangothini lwesitoreji samanje.
Ingxenye 1: Ukukholwa EThebulani
Ukuze senze usekelo lwe-IT lube lula futhi senze izinguquko ngokushesha, sinqume ukwenza ingqondo yokubala izinkomba ezingangezeleli nokuqhathanisa izikhathi ezidlule ku-Tableau.
Isiteji 1. Yonke into Ibukhoma, akukho ukuguqulwa kwamafasitela.
Kulesi sigaba, sixhume i-Tableau ngaphambili kwezitolo zamanje futhi sanquma ukubona ukuthi inani lamaresidi onyaka owodwa lizobalwa kanjani.
Umphumela:
Impendulo yayicindezela - imizuzu engu-20. Ukudluliswa kwedatha ngenethiwekhi, umthwalo ophezulu ku-Tableau. Sibonile ukuthi i-logic enezinkomba ezingezona ezengezo idinga ukusetshenziswa ku-HANA. Lokhu akuzange kusesabise kakhulu, sesivele sinesipiliyoni esifanayo ne-BO nokuhlaziya futhi sasazi ukuthi singakha kanjani imibukiso esheshayo ku-HANA ekhiqiza izinkomba ezingezona ezengezo ezibalwe kahle. Manje osekusele nje wukuwalungisela eTableau.
Isiteji sesi-2. Sishuna amakesi okubonisa, akukho ukwenziwa kwezinto, yonke into ngokundiza.
Sidale umbukiso omusha ohlukile okhiqize idatha edingekayo ye-TABLEAU ngokuphazima kweso. Ngokuvamile, sithole umphumela omuhle; sinciphise isikhathi sokukhiqiza zonke izinkomba ngeviki elilodwa ukuya kumasekhondi angu-9-10. Futhi besilindele ngobuqotho ukuthi ku-Tableau isikhathi sokuphendula sedeshibhodi sizoba imizuzwana engu-20-30 ekuvuleni kokuqala futhi ngenxa yenqolobane kusuka ku-10 kuya ku-12, okuzovumelana nathi ngokujwayelekile.
Umphumela:
Ideshibhodi yokuqala evuliwe: imizuzu emi-4-5
Noma yikuphi ukuchofoza: 3-4 amaminithi
Akekho owayelindele ukwanda okwengeziwe okunjalo emsebenzini wesitolo sangaphambili.
Ingxenye 2. Gxumela ku-Tableau
Isiteji 1. Ukuhlaziywa kokusebenza kwethebula nokulungisa okusheshayo
Siqale ukuhlaziya lapho iTableau ichitha khona isikhathi sayo esiningi. Futhi kukhona amathuluzi amahle kakhulu alokhu, okuyinto, vele, i-plus ye-Tableau. Inkinga enkulu esiyitholile kwakuyimibuzo eyinkimbinkimbi ye-SQL eyayakha i-Tableau. Ngokuyinhloko zazihlotshaniswa nalokhu:
- ukuguqulwa kwedatha. Njengoba i-Tableau ingenawo amathuluzi okudlulisa amasethi edatha, ukuze sakhe uhlangothi lwesobunxele lwedeshibhodi enokumelwa okuningiliziwe kwawo wonke ama-KPI, kudingeke ukuthi sakhe ithebula sisebenzisa ikesi. Usayizi wemibuzo ye-SQL kusizindalwazi ufinyelele izinhlamvu ezingu-120.
- ukukhetha isikhathi. Umbuzo onjalo ezingeni lesizindalwazi uthathe isikhathi esiningi ukuhlanganisa kunokusebenzisa:
Labo. cela ukucubungula imizuzwana eyi-12 + ukubulawa kwemizuzwana emi-5.
Sinqume ukwenza lula i-logic yokubala ohlangothini lwe-Tableau futhi sihambise enye ingxenye yezibalo ezingeni lesitolo sangaphambili kanye ne-database. Lokhu kwaletha imiphumela emihle.
Okokuqala, senze ukuguqulwa empukaneni, sikwenze ngokujoyina okuphelele kwangaphandle esigabeni sokugcina sokubala kwe-VIEW, ngokwale ndlela echazwe ku-wiki. ΠΈ .
Okusho ukuthi, senze itafula lokusetha - i-matrix yokuguqula (21x21) futhi sathola zonke izinkomba ngokulandelana komugqa nomugqa.
Bekuyi:
Kwaba:
Cishe asikho isikhathi esichithwa ku-database transposition ngokwayo. Isicelo sazo zonke izinkomba zesonto siqhubekile nokusetshenzwa ngemizuzwana eyi-10. Kodwa ngakolunye uhlangothi, ukuguquguquka kuye kwalahleka mayelana nokwakhiwa kwedeshibhodi ngokusekelwe kunkomba ethile, i.e. ohlangothini lwesokudla lwedeshibhodi, lapho okuguquguqukayo kanye nokuhlukaniswa okuningiliziwe kwenkomba ethile kuvezwa, ngaphambilini iwindi lokubonisa lalisebenza ngemizuzwana engu-1-3, ngoba isicelo sasisekelwe kunkomba eyodwa, futhi manje i-database ihlale ikhetha zonke izinkomba futhi ihlunga umphumela ngaphambi kokubuyisela umphumela ku-Tableau.
Ngenxa yalokho, ijubane ledeshibhodi lehle cishe izikhathi ezi-3.
Umphumela:
- 5 isekhondi β amadeshibhodi okuhlaziya, okubonwayo
- Imizuzwana engu-15-20 - ukulungiselela ukuhlanganisa imibuzo ngokwenza izibalo zangaphambili ku-Tableau
- 35-45 sec - ukuhlanganiswa kwemibuzo ye-SQL kanye nokusebenza kwayo okulandelanayo okulandelanayo e-Hana
- 5 amasekhondi β kucutshungulwa imiphumela, ukuhlela, ukubala kabusha ukubonwa ku-Tableau
- Yiqiniso, imiphumela enjalo ayizange ihambisane nebhizinisi, futhi saqhubeka nokwenza kahle.
Isigaba 2. I-logic encane ku-Tableau, ukwenziwa kwezinto ezibonakalayo okuphelele
Saqonda ukuthi kwakungenakwenzeka ukwakha ideshibhodi enesikhathi sokuphendula samasekhondi ambalwa endaweni yesitolo esebenza imizuzwana engu-10, futhi sicabangele izinketho zokwenza idatha kuhlangothi lwedathabheyisi ngokuqondile ideshibhodi edingekayo. Kodwa sihlangabezane nenkinga yomhlaba wonke echazwe ngenhla - izinkomba ezingangezeleli. Asikwazanga ukwenza isiqiniseko sokuthi lapho sishintsha izihlungi noma ama-drilldowns, i-Tableau ishintsha kalula phakathi kwangaphambili kwesitolo namazinga adizayinelwe kusengaphambili izigaba zemikhiqizo ehlukene (ngokwesibonelo, imibuzo emithathu ngaphandle kwe-UTE, ne-UTE1 ne-UTE2 ikhiqiza imiphumela ehlukene). Ngakho-ke, sinqume ukwenza lula ideshibhodi, silahle ukulandelana komkhiqizo kudeshibhodi futhi sibone ukuthi kungashesha kangakanani enguqulweni eyenziwe lula.
Ngakho-ke, kulesi sigaba sokugcina, sihlanganise inqolobane ehlukile lapho sengeze khona wonke ama-KPI ngendlela eguquliwe. Ohlangothini lwedathabheyisi, noma yisiphi isicelo endaweni enjalo sicutshungulwa ngemizuzwana engu-0,1 - 0,3. Kudeshibhodi sithole imiphumela elandelayo:
Ukuvulwa kokuqala: imizuzwana engu-8-10
Noma yikuphi ukuchofoza: imizuzwana engu-6-7
Isikhathi esichithwa yi-Tableau siqukethe:
- 0,3 isekhondi. - Ukuhlukaniswa kwedeshibhodi kanye nokuhlanganiswa kwemibuzo ye-SQL
- 1,5-3 isekhondi. - ukusetshenziswa kwemibuzo ye-SQL ku-Hana ukuze kubonakale okuyinhloko (igijima ngokuhambisana nesinyathelo 1)
- 1,5-2 isekhondi. - ukunikeza, ukubalwa kabusha kokubonwayo
- 1,3 isekhondi. - Ukwenziwa kwemibuzo eyengeziwe ye-SQL ukuze uthole amanani okuhlunga afanelekile (Ibhrendi, Isigaba, Idolobha, Isitolo), imiphumela yokuhlaziya
Ukukufingqa kafushane
Sithande ithuluzi le-Tableau ngokombono wokubuka. Esigabeni se-prototyping, sicabangele izici ezihlukahlukene zokubuka futhi sazithola zonke kumalabhulali, okuhlanganisa ukuhlukaniswa okuyinkimbinkimbi kwamaleveli amaningi kanye nempophoma enabashayeli abaningi.
Ngenkathi sisebenzisa amadeshibhodi anezinkomba zokuthengisa ezibalulekile, sihlangabezane nezinkinga zokusebenza esingakakwazi ukuzinqoba. Sichithe isikhathi esingaphezu kwezinyanga ezimbili futhi sathola ideshibhodi engaphelele engasebenzi, isivinini sokuphendula esisonqenqemeni lokwamukeleka. Futhi sizifinyelelele iziphetho:
- I-Tableau ayikwazi ukusebenza ngenani elikhulu ledatha. Uma kumodeli yedatha yasekuqaleni unedatha engaphezu kuka-10 GB (imigqa elinganiselwa ezigidini ezingu-200 X 50), ideshibhodi izonciphisa kakhulu ijubane - ukusuka kumasekhondi angu-10 ukuya emaminithini ambalwa ngokuchofoza ngakunye. Sizame kokubili ukuxhumana okubukhoma nokukhipha. Isivinini sokusebenza siyaqhathaniswa.
- Umkhawulo lapho usebenzisa isitoreji esiningi (amadathasethi). Ayikho indlela yokubonisa ubudlelwano phakathi kwamasethi edatha usebenzisa izindlela ezijwayelekile. Uma usebenzisa ama-workaround ukuze uxhume amasethi edatha, lokhu kuzoba nomthelela omkhulu ekusebenzeni. Esimweni sethu, sicabangele inketho yokwenza idatha ibe ngokoqobo esigabeni ngasinye sokubuka esidingekayo nokwenza izinguquko kulawa madathasethi enziwe izinto kuyilapho silondoloza izihlungi ezikhethwe ngaphambili - lokhu kubonakale kungenakwenzeka ku-Tableau.
- Akwenzeki ukwenza amapharamitha aguqukayo ku-Tableau. Awukwazi ukugcwalisa ipharamitha esetshenziselwa ukuhlunga idathasethi kusihumusho noma phakathi nokuxhuma bukhoma ngomphumela wenye ukukhethwa kudathasethi noma umphumela womunye umbuzo we-SQL, okokufaka komsebenzisi womdabu kuphela noma okungaguquki.
- Imikhawulo ehlobene nokwakha ideshibhodi enezici ze-OLAP|PivotTable.
Ku-MSTR, SAP SAC, SAP Analysis, uma wengeza idathasethi embikweni, khona-ke zonke izinto ezikuyo zihlobene ngokuzenzakalelayo. I-Tableau ayinakho lokhu; uxhumano kufanele lulungiswe ngesandla. Lokhu cishe kuvumelana nezimo, kodwa kuwo wonke amadeshibhodi ethu lokhu kuyisidingo esiyisibopho sezakhi - ngakho-ke lezi yizindleko zabasebenzi ezengeziwe. Ngaphezu kwalokho, uma wenza izihlungi ezihlobene ukuze, ngokwesibonelo, lapho uhlunga isifunda, uhlu lwamadolobha lukhawulelwe emadolobheni alesi sifunda kuphela, ugcina unemibuzo elandelanayo ku-database noma i-Extract, ebambezela ngokuphawulekayo ideshibhodi. - Imikhawulo emisebenzini. Uguquko oluningi alunakwenziwa ekukhishweni noma, IKHAYA KAKHULU, kudathasethi evela ku-Live-connecta. Lokhu kungenziwa nge-Tableau Prep, kodwa kuwumsebenzi owengeziwe kanye nelinye ithuluzi lokufunda nokuligcina. Isibonelo, awukwazi ukudlulisa idatha noma ukuyijoyina ngokwayo. Okuvalwa ngokuguqulwa kwamakholomu angawodwana noma izinkambu, okufanele zikhethwe ngokwesibonelo noma uma, futhi lokhu kudala imibuzo eyinkimbinkimbi ye-SQL, lapho isizindalwazi sichitha khona isikhathi saso esiningi sihlanganisa umbhalo wombuzo. Lokhu kuguquguquka kwethuluzi kwadingeka kuxazululwe ezingeni lokubonisa, okuholela ekugcinweni okuyinkimbinkimbi, ukulandwa okwengeziwe nokuguqulwa.
Asikalilahli ithemba nge-Tableau. Kodwa asithathi i-Tableau njengethuluzi elikwazi ukwakha amadeshibhodi ezimboni kanye nethuluzi lokushintsha nokwenza idijithali lonke uhlelo lokubika lwenkampani yenkampani.
Manje sithuthukisa ideshibhodi efanayo kwelinye ithuluzi futhi, ngesikhathi esifanayo, sizama ukubuyekeza ukwakheka kwedeshibhodi ku-Tableau ukuze siyenze ibe lula nakakhulu. Uma umphakathi unentshisekelo, sizokutshela ngemiphumela.
Siphinde silinde imibono noma izeluleko zakho zokuthi e-Tabeau ungawakha kanjani amadeshibhodi asheshayo phezu kwedatha enkulu kangaka, ngoba sinewebhusayithi lapho kukhona idatha eningi kakhulu kuneyezitolo.
Source: www.habr.com