Sensors and network administering


Data Mining And Modelling

The important processes that have to be
clearly delineated for Data Mining, AnalysisMuch of the data that they have will have
and  Modelling  are:different frequencies of change, refreshment
or occurrence. It will be kept for different
Data model: what data will be available andperiods. In some cases, aggregated data may
how  will  it  flow?be kept rather than source data. All of these
factors effect the data modelling exercise
Data gathering: how will data be gatheredand the eventual modelling software
both  in  physical  and  technological terms?requirements.
Data  gathered:  what  data will be gathered?Turning the data into useful information
requires:
Data types: what types of data will be
gathered?Identifying  the  issue(s)
Data  formatting:  how  will  data  be  held?Assembling  the  data  set(s)
Data  warehousing:  where  will data be held?Building  models
Data mining: how will we retrieve data fromVerify  models
the  warehouse?
Interpretation  of  the  results
Information modelling: how will we create
models  and  what  of?Automation  of  the  delivery
Information access: how will we access theThereafter, modelling tools and techniques
data  models  and  reports?have to be used. These can be divided into
two  groups:  theory  driven and data driven.
Presentation & reporting: on what will we
report?Theory driven modelling (hypothesis testing)
attempts to substantiate or disprove
Most companies want to know essentialpreconceived ideas. Theory driven modelling
information about customers at every point oftools require the user to specify most of the
contact,  for  example:model based on prior knowledge and then tests
to  see  if  the  model  is  valid.
Lifetime  value
Data driven modelling tools automatically
X  sell  and  upgrade  potentialcreate the model based on patterns they find
in the data. This also needs to be tested
Acquisition  costbefore  it  can  be  accepted  as  valid.
Channel  preferencesModelling is an iterative process with the
final model usually being a combination of
Loyalty/retentionprior knowledge and newly discovered
information.
Purchase  behaviour  patterns



1 A B C 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95