Data Analytics Palast | | §-cem-c ASsiqnment-ot Day Ace Tt Explain the Data Architecture fox Desig - % Data Architecture ‘Ts composed of models, Policies, tules and “how 01 Standards tat qovem which data is collected} e cht Mic stored, cevvanged, integrate and put ‘to us | System and in organization s for ures IP Data Architecture’ ts Pichitect is desponsible | data ~ standing busines objectives and the exicting infiashuctuve and) assets. fe Data Archifeckse design is set of Stand ands whieh | Cowpered OF cegtuin policies, ules, models Ke Data Prebitecture also desenbet the type at lata -to Manage data .and Ptovide san) Shire tures applied easy wal for data pre proce ving * Pacts thancing deta fnditechre:— G) entexprise cequiveinent — tt thcludes expansion of business if pexfexrmance of syle data Management shansaction Management ft Making use of raw data by commenting thet into Use -fal Rles Storing of data in data Warehouse , accesNE Conon) CS w eer s “ These are ako fnportant Pretoas thed must be consida 7 ¥ duaing the data Prchiteclune Phese , while opkmal ‘o a is possitle thal Some solutions, ales dae th" Painciple, may not be political candid cost ~ it) Businece policies: prahilecture desi Rusines policter that alto duive deco. Include * Pnteinal organizational poltexes | bodies | X rales of equlaton % — profexnenal ‘Stemdards. that can be vay applicable agency). Pacbitectune at three by Designing the tevele af SPecbiodten * The logteal level % he poysicl level X The Drrplementatin level see Models " | eS + Pe . [ “Teckricad — }-— [er LA i aequivcment A ae Hi ea —, L_Modelting Phytic) ee“ ‘| date Male? -equivernen's | / rT) Business bate | i Gi =e ee clieeialeveleiea cali benta ues nies it ekMaley The The loyal view | usess View : a Data finelytics teptesents ata in a format that Ts a User and +o the progans thal “rou irveccina sl bo bose data Logical View tells the Uses, twhat ts dotabese , Logical levell coneicts c@ data wcequivements and pretew data mnodeling modele whtch are proce ed tasing fechriquer to recut: tn logical data eno del Phyi cal level?— x pt is cxeat W phytical Jala, Wo the database Khe Medd ic created by the Soflusave architects, slo develope’ 4o the level Fem logical level and various used here with inpud dint nis patos! Rom software developes of datat dlata todelling qechniques ave Vanous Roswnalls ed when we Hanclale the top \evel design) date bare cadnite ck, 4 of Aatabase Ddminihatos _ by The Input data modelling fechriquesare Ie “These ot xepeerentation of Nala such a8 relat eal deta model, aetworle wodd bictotorl mode eotty nelohents woe, v Enplementatien, level pacataten aS BE contains details about modiGeahen and ot various date mining 1 cat deta. Aryoegh tne Use Such as (2-stedie, ween, Gtarge ete) (F Reve each tool tet a specific feakuve be ib was ArPresond cept: sy OF Ni Sy an Jee exntalye viewing aiso method !— a % The obsewation method ic a methed of Aata colleetiso| Tm uhich the ceccarches ceenly obtcaves the behavieu cand practice, Sf the engek audience Using some dad Collecting 4ovl and cioses the obeaved aaa in the fon of 4exl, audio, vide ov any ow format » ; FM thic melted, the data collected diveclty by partici p ant Posting a fess question on the % he data obtained wil be cent fos processing) 1) Sunvey methods the procew of Texearch ushere and antwe eK The Suwey method is a lick of elevant JYertene ave aglce. Save note) dawn in the form of text audio ot vi methed Can be obtained ty both) line The sues { alsline mode . g Like through website Frome and ermal, then tak | ove fox analyzing date : Suwey anwes aye © eompls ave eoline suey ef Sumeys “through J eial media pale - a : 4 E fost vii o> ee eycy Pe most | Ficque nth “used expennent methode one 1a Completely Randomized des 10 Randomized Blode derign Wi) Latin square design Ww) Fachwal desigo Sources of Secondary Deda + Secon das data can be obtained Brough. ¥ Dylernal courced — within the otganisebien + Sukerna) Sources - outste the esgantsakien E)) Prieryal Sources = oblained with lees time, includes HPL may) be effosts, noncy FX Thermal Sources oO Accoun ting resources Gy Sales force Repent wi) Botemal Fxperts () Miscellancous Reports ata t=labouy Bureau Depaotment of eesnomic foie “ Sinte statistical Abshack Non Government puiblicdtiong . Qyndicade eegivicas - Explain goigy outliers micting Valucs - Notsy data en Missing data is Meaningle es deta , * ‘Thic inoludes data Comuuption and the ferm is oHen Used ag a synonym t% comupt data that & user Surtery ensiech . ¥% Te also indude any data interprated Cannot tendertand and The following ave the technique help to remove oni: ‘Binning es > Binnio meltode Smouth a Stoved data value hy consulting Hs “neighbourhood ” sthad is the value ago it. ; he sevted values ave Aichibuted info 00 .oFBeaass® Pinning metho dy “conoult he ) AD - perform, Nocal tenoothing - ; [* M Sroosthing by bin teans, each value in ou is niepla ced by the teat value of the bin. He To Smoothing is neplaced by the median Valte the Mines) bin boundan’ ey , o tdentified at] TB smooth nl by and madmum Values is given bin an she din bourlawes - Outliers - ; that baoimal % An outlier ic an obsegvation fer an 4 Pltetaeee hem cihes Values a a random sarepe QA poprlati«n Me gutters can be clastified toto tyee categories t= Gi) contextual outties UD) qlobal outle (48) caltectve outer - 1) Global oullies Nobal_oullis outlier (F Mm a gyver dataset, Jata object is % glove! Abt deviates significantly from the vext- of the data . | Global outliers ane gometimes Called peint aroma i i i i andi ave the ingles tyPe OF outers,ite j en Q) conkextual alliers:~ a. a data objet is a cork ¥ Toa given data set, Outlier if it deviate Signi freantly wilh Teapect to a \ Speci Pie context of the object © Contextual outliers are alto lunown ac condchonal auttien becauye they are conditional on the Selcde. Context - 2) collective outlier !~” pobseivey ou % Th a dataset, % Subtet of data objects Rom | outties if the objec a2 data ech ushole a collec Hve deviate Signi Realty som the * STuportanty , the individual data objects ry not be outtien . eg tine Missing values :— T ts very much useful to have roi eting value in yo data set te tony have hapenea) dosing dala collection Estinating mows with mictirg data Estimate misting vabies aceie the duple Fillig the roicting value ‘anna vse 2 global conctaxt ty fill in the misting value |, Use a tmeasuve of central Aenden eat fox. the Heawastt Ontable Valuto to Fill dhe mitetrs ValieoPonte Yown any 6 tools Wp Gato-Pinalytics 9 | a ) Repregztarming 1 1" 4 YR is the most used Prograroming language fox developing [Statistical tools: TM can easily manipulate yout data and present \p }) different ways R compiles and nuns on vamious Platform guch og Unix, wirdews and mac os SR prvide, vast number Of Pacleages and buil 19 Pamptions whids can be automatically install R provides quality letting and pephiog Gi) Prythoo t= Python is an openn Source, object oriented progann\ a language which, is easy ‘to read, write and minted 4 TE way developed by Guido van Rortum in late: Iqads which Suppers both Fanetiona) and cul Programming mettod ¢- Th peed leas Winer of code te pevfvrm the same tose og Sd to athe foo lanquag’s Wee | ‘ole3) Che) Refine: gf ¥ Neo known 3 Google Refine . e ¥ This data cleaning sofkoare will help you cleant , up data for analysis \ Te is used for cleaning meety data, te Hantfoing, wH data and pasting dala From webster. 4) Rapid Mines! DD TWic too] is mosthy ured for predictive analytics Such ag data mining, text aeunalytis, Machene (earning And vicval analytics worthout any programming A paserfal, intaiatel plattown that can \nteqralle L | with ane data soumee tyres Such ay Accom, Excel) | Macle, Let ele p> hic tool is ver proertal that can qeneroteonalytis bared on real-life dala tranctormation sing ‘v€ | You can contol the farmats ant data eek fos. Predictve analytis |5) Ppache spate :— ~) The University of california, Bereelags Amp Lab, developed Apache in 2009 - 5 at isa fast lasge- sete data pro cesting -exgina | = Ths tor\ executes applications in aulters lov times Frasher 1) Memery and to me Fe2t® on dite > my & Bult en Zeenat iag nothing but a process trough ioAelting i 2p a format 10 a databas 4 of deka Models ! — r= ievarchieal Model " oy I) is the Name *ndicates, this model maleer Nem Keine +o Shuclure netneving \he dato in o Aveo Wee Porm Mowever , and accessing data is Aifeult Wy Avie Wodel . f athich hos, cont Ae 2 tree odd clants Foor the wo Ahe heinarchy expands in the Form of 2 aml then it child node to the Parent node - S This model is cacy 40 feprerent Some toor\d aclation thin (82) Relational Mole| 7= 5 data 1S represented 5 Ali the ‘Information ty Shred in the Sem & and columns: ae Ye ‘ b weduces complexity el provides 0 iclex of the Qata J Tk ic a Simple Model jlo at the 2! Wn the form ot tables -(It) Nehumfe Model! ft Ss oh The nelumk model is extention of the heevarchical 1 tnodel- =) However, unlike the hierarchiad mod! lex relation eho pt 03 th multiple parent yecowdy a), thie troded Males it easies to Comey ¢ each tecord can be linked tot Here a child node could have ) Navigation it Paster, ag there are multirle paths to wench a child entity) - W) object - ome ated mode lt— J This model consists of a collection at objects, with ite features and methods this type of model te alco called the Pott ges | Aatabase model ' To Uhis model, fue ave more objects thy ough links multiple paver nodes each ane cornedel) @) Eokly - aelakonehip Model * Enbty ~ eladiondhip model alee known a ee Mosel { Sepregents entke, and ther eelockion ship 1 a qeaphi format L, Ay entity could be anything ~ 4 emer, a Piecot data os oS aed te te <984 4 build . MP we lenow the relation hip belweon the ctnbuter and entities. |