Session 4-6: Business Intelligence !esol"ing the Pu##le$
Business Intelligence resol"ing the %u##le$ &hat is a decision ' &hat are the (inds o) decisions you ma(e in your li)e ' *aily ' &ee(ly ' Monthly ' +early ' In a li)etime ' ,et-s hear some e.am%les/
Business Intelligence resol"ing the %u##le$ &hich o) these decisions 0ould you say are -tactical- and 0hich are -strategic- ' &hy '
Business Intelligence resol"ing the %u##le$ Is it easier to ma(e a -tactical- decision or to ma(e a -strategic- decisions in li)e ' &hich ha"e a longer im%act ' &hich ta(e longer to decide ' &hich ta(e longer to im%lement ' &hich consume more resources 1money, time, e))ort, ///2 to im%lement '
Business Intelligence resol"ing the %u##le$ &hat do you do 3e)ore ma(ing a -strategic- decision '
Business Intelligence resol"ing the %u##le$ Be)ore ma(ing a strategic decision, 0e: *e"elo% 1)easi3le2 alternati"es 1select a )e0 0e li(e2/ Collect data/ Clean it 1"eri)y, cross chec(, )ill ga%s, remo"e unnecessary, etc2/ 4rganise and store it/ *ata analysis 1statistics2/ 5%%ly learning 1in)erences2 to our decision ma(ing/
Business Intelligence resol"ing the %u##le$ Is data a"aila3le al0ays structured data ' &hat is unstructured data 1Big *ata2 ' &hat are the sources o) Big *ata '
Business Intelligence resol"ing the %u##le$ &hat i): ,imited data is a"aila3le/ ,imited time is a"aila3le/ ,imited (no0ledge )or analysis is a"aila3le' ,imited %rocessing %o0er )or analysis is a"aila3le' 5nd, data to analyse is huge $
Business Intelligence resol"ing the %u##le$ Can you cite e.am%les o) -decision ma(ing in certainty- )rom your li)e ' 6o0 a3out -decision ma(ing in uncertainty- ' 1Multi%le choice 7uestions 7ui# 0ith 8 0ithout negati"e mar(s$2
Business Intelligence resol"ing the %u##le$ *o 0e al0ays ha"e all needed data and analytical a3ility to ma(e structured decisions ' &hen you are %laying ta3le tennis "ersus 0hen you are %laying chess ' &hich is a s(ill 3ased game ' &hich is a strategy 3ased game ' &hy '
Business Intelligence resol"ing the %u##le$ Situations 0hich managers )ace: Structured and unstructured data/ Incom%lete$ ,imited time, resources, a3ility 19:S2 at hands/ Modern tools ' Can they decide )or you ' Can they re%lace a manager '
Business Intelligence resol"ing the %u##le$ &hat is a data 0arehouse ' &hat is data mining ' Business Intelligence 19no0ledge disco"ery in data3ases, 9**2'
Business Intelligence resol"ing the %u##le$ *ata 0arehouses store historic data/ The storage may 3e aggregated i) analysis does not demand atomic le"el data/ ; This ma(es %rocessing8analysis )aster/ 6o0e"er, granularity is lost/ The storage may 3e de-normalised/ ; Increases s%ace ta(en to store, 3ut reduces %rocessing time 1a s%ace time trade o))2/
Business Intelligence resol"ing the %u##le$ Managers should esta3lish the goals 1Business o3<ecti"es2 clearly 3e)ore "enturing into *&, *M, 9**, BI/ 5lmost al0ays, there are %re"iously un(no0n 3ene)its o) BI $$ Be a0are o) statisticians 0ho ma(e tall claims, such as: 6al) o) India-s 0or(ing %o%ulation earns less than the median salary/ ,atest sur"ey sho0s that = out o) 4 %eo%le ma(e u% >?@ o) the 0orld-s %o%ulation/
Business Intelligence resol"ing the %u##le$ Managers must also use commonly a"aila3le statistical tools 3e)ore "enturing into an e.%ensi"e detailed analysis o) large data/ Single dimensional data aggregation: a"erage 8 mean, median, mode, %ercentile /// etc/ T0o dimensional: gra%hs, %ie charts, etc/ Three dimensional using solid models or modern com%uters: data cu3es 1slice and dice data2, data "isualisation, etc/ Ty%ical o%erations on data: Classi)ication, Clustering 1and outlier analysis2, 5ssociation rule, Pattern 1se7uence2 disco"ery/
Business Intelligence resol"ing the %u##le$ Best o) luc( )or the 7ui# $$ Ae.t class: the readings 1and *M, *&, BI2 in more detail, im%act o) BI on 3usinesses/